1 \input texinfo @c -*-texinfo-*-
2 @comment %**start of header
3 @setfilename ../../info/eintr
4 @c setfilename emacs-lisp-intro.info
5 @c sethtmlfilename emacs-lisp-intro.html
6 @settitle Programming in Emacs Lisp
11 @include emacsver.texi
13 @c ================ How to Print a Book in Various Sizes ================
15 @c This book can be printed in any of three different sizes.
16 @c Set the following @-commands appropriately.
26 @c European A4 size paper:
31 @c (Note: if you edit the book so as to change the length of the
32 @c table of contents, you may have to change the value of `pageno' below.)
34 @c <<<< For hard copy printing, this file is now
35 @c set for smallbook, which works for all sizes
36 @c of paper, and with PostScript figures >>>>
44 @c ================ Included Figures ================
46 @c If you clear this, the figures will be printed as ASCII diagrams
47 @c rather than PostScript/PDF.
48 @c (This is not relevant to Info, since Info only handles ASCII.)
49 @set print-postscript-figures
50 @c clear print-postscript-figures
52 @comment %**end of header
54 @c per rms and peterb, use 10pt fonts for the main text, mostly to
55 @c save on paper cost.
56 @c Do this inside @tex for now, so current makeinfo does not complain.
62 \global\hbadness=6666 % don't worry about not-too-underfull boxes
65 @c These refer to the printed book sold by the FSF.
66 @set edition-number 3.10
67 @set update-date 28 October 2009
69 @c For next or subsequent edition:
70 @c create function using with-output-to-temp-buffer
71 @c create a major mode, with keymaps
72 @c run an asynchronous process, like grep or diff
74 @c For 8.5 by 11 inch format: do not use such a small amount of
75 @c whitespace between paragraphs as smallbook format
78 \global\parskip 6pt plus 1pt
82 @c For all sized formats: print within-book cross
83 @c reference with ``...'' rather than [...]
85 @c This works with the texinfo.tex file, version 2003-05-04.08,
86 @c in the Texinfo version 4.6 of the 2003 Jun 13 distribution.
89 \if \xrefprintnodename
90 \global\def\xrefprintnodename#1{\unskip, ``#1''}
92 \global\def\xrefprintnodename#1{ ``#1''}
94 % \global\def\xrefprintnodename#1{, ``#1''}
97 @c ----------------------------------------------------
99 @dircategory Emacs lisp
101 * Emacs Lisp Intro: (eintr). A simple introduction to Emacs Lisp programming.
105 This is an @cite{Introduction to Programming in Emacs Lisp}, for
106 people who are not programmers.
109 Edition @value{edition-number}, @value{update-date}
112 Distributed with Emacs version @value{EMACSVER}.
115 Copyright @copyright{} 1990--1995, 1997, 2001--2013 Free Software
122 GNU Press, @hfill @uref{http://www.fsf.org/licensing/gnu-press/}@*
123 a division of the @hfill email: @email{sales@@fsf.org}@*
124 Free Software Foundation, Inc. @hfill Tel: +1 (617) 542-5942@*
125 51 Franklin Street, Fifth Floor @hfill Fax: +1 (617) 542-2652@*
126 Boston, MA 02110-1301 USA
130 Printed copies available from @uref{http://shop.fsf.org/}. Published by:
133 GNU Press, http://www.fsf.org/licensing/gnu-press/
134 a division of the email: sales@@fsf.org
135 Free Software Foundation, Inc. Tel: +1 (617) 542-5942
136 51 Franklin Street, Fifth Floor Fax: +1 (617) 542-2652
137 Boston, MA 02110-1301 USA
145 Permission is granted to copy, distribute and/or modify this document
146 under the terms of the GNU Free Documentation License, Version 1.3 or
147 any later version published by the Free Software Foundation; there
148 being no Invariant Section, with the Front-Cover Texts being ``A GNU
149 Manual'', and with the Back-Cover Texts as in (a) below. A copy of
150 the license is included in the section entitled ``GNU Free
151 Documentation License''.
153 (a) The FSF's Back-Cover Text is: ``You have the freedom to
154 copy and modify this GNU manual. Buying copies from the FSF
155 supports it in developing GNU and promoting software freedom.''
159 @c half title; two lines here, so do not use `shorttitlepage'
162 \hbox{}\vskip 1.5in \chaprm \centerline{An Introduction to}%
164 {\begingroup\hbox{}\vskip 0.25in \chaprm%
165 \centerline{Programming in Emacs Lisp}%
166 \endgroup\page\hbox{}\page}
171 @center @titlefont{An Introduction to}
173 @center @titlefont{Programming in Emacs Lisp}
175 @center Revised Third Edition
177 @center by Robert J. Chassell
180 @vskip 0pt plus 1filll
186 @evenheading @thispage @| @| @thischapter
187 @oddheading @thissection @| @| @thispage
191 @c Keep T.O.C. short by tightening up for largebook
194 \global\parskip 2pt plus 1pt
195 \global\advance\baselineskip by -1pt
205 @top An Introduction to Programming in Emacs Lisp
209 <p>The homepage for GNU Emacs is at
210 <a href="/software/emacs/">http://www.gnu.org/software/emacs/</a>.<br>
211 To view this manual in other formats, click
212 <a href="/software/emacs/manual/eintr.html">here</a>.
218 This master menu first lists each chapter and index; then it lists
219 every node in every chapter.
222 @c >>>> Set pageno appropriately <<<<
224 @c The first page of the Preface is a roman numeral; it is the first
225 @c right handed page after the Table of Contents; hence the following
226 @c setting must be for an odd negative number.
229 @c global@pageno = -11
232 @set COUNT-WORDS count-words-example
233 @c Length of variable name chosen so that things still line up when expanded.
236 * Preface:: What to look for.
237 * List Processing:: What is Lisp?
238 * Practicing Evaluation:: Running several programs.
239 * Writing Defuns:: How to write function definitions.
240 * Buffer Walk Through:: Exploring a few buffer-related functions.
241 * More Complex:: A few, even more complex functions.
242 * Narrowing & Widening:: Restricting your and Emacs attention to
244 * car cdr & cons:: Fundamental functions in Lisp.
245 * Cutting & Storing Text:: Removing text and saving it.
246 * List Implementation:: How lists are implemented in the computer.
247 * Yanking:: Pasting stored text.
248 * Loops & Recursion:: How to repeat a process.
249 * Regexp Search:: Regular expression searches.
250 * Counting Words:: A review of repetition and regexps.
251 * Words in a defun:: Counting words in a @code{defun}.
252 * Readying a Graph:: A prototype graph printing function.
253 * Emacs Initialization:: How to write a @file{.emacs} file.
254 * Debugging:: How to run the Emacs Lisp debuggers.
255 * Conclusion:: Now you have the basics.
256 * the-the:: An appendix: how to find reduplicated words.
257 * Kill Ring:: An appendix: how the kill ring works.
258 * Full Graph:: How to create a graph with labeled axes.
259 * Free Software and Free Manuals::
260 * GNU Free Documentation License::
265 --- The Detailed Node Listing ---
269 * Why:: Why learn Emacs Lisp?
270 * On Reading this Text:: Read, gain familiarity, pick up habits....
271 * Who You Are:: For whom this is written.
273 * Note for Novices:: You can read this as a novice.
278 * Lisp Lists:: What are lists?
279 * Run a Program:: Any list in Lisp is a program ready to run.
280 * Making Errors:: Generating an error message.
281 * Names & Definitions:: Names of symbols and function definitions.
282 * Lisp Interpreter:: What the Lisp interpreter does.
283 * Evaluation:: Running a program.
284 * Variables:: Returning a value from a variable.
285 * Arguments:: Passing information to a function.
286 * set & setq:: Setting the value of a variable.
287 * Summary:: The major points.
288 * Error Message Exercises::
292 * Numbers Lists:: List have numbers, other lists, in them.
293 * Lisp Atoms:: Elemental entities.
294 * Whitespace in Lists:: Formatting lists to be readable.
295 * Typing Lists:: How GNU Emacs helps you type lists.
299 * Complications:: Variables, Special forms, Lists within.
300 * Byte Compiling:: Specially processing code for speed.
304 * How the Interpreter Acts:: Returns and Side Effects...
305 * Evaluating Inner Lists:: Lists within lists...
309 * fill-column Example::
310 * Void Function:: The error message for a symbol
312 * Void Variable:: The error message for a symbol without a value.
316 * Data types:: Types of data passed to a function.
317 * Args as Variable or List:: An argument can be the value
318 of a variable or list.
319 * Variable Number of Arguments:: Some functions may take a
320 variable number of arguments.
321 * Wrong Type of Argument:: Passing an argument of the wrong type
323 * message:: A useful function for sending messages.
325 Setting the Value of a Variable
327 * Using set:: Setting values.
328 * Using setq:: Setting a quoted value.
329 * Counting:: Using @code{setq} to count.
331 Practicing Evaluation
333 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
335 * Buffer Names:: Buffers and files are different.
336 * Getting Buffers:: Getting a buffer itself, not merely its name.
337 * Switching Buffers:: How to change to another buffer.
338 * Buffer Size & Locations:: Where point is located and the size of
340 * Evaluation Exercise::
342 How To Write Function Definitions
344 * Primitive Functions::
345 * defun:: The @code{defun} macro.
346 * Install:: Install a function definition.
347 * Interactive:: Making a function interactive.
348 * Interactive Options:: Different options for @code{interactive}.
349 * Permanent Installation:: Installing code permanently.
350 * let:: Creating and initializing local variables.
352 * else:: If--then--else expressions.
353 * Truth & Falsehood:: What Lisp considers false and true.
354 * save-excursion:: Keeping track of point, mark, and buffer.
358 Install a Function Definition
360 * Effect of installation::
361 * Change a defun:: How to change a function definition.
363 Make a Function Interactive
365 * Interactive multiply-by-seven:: An overview.
366 * multiply-by-seven in detail:: The interactive version.
370 * Prevent confusion::
371 * Parts of let Expression::
372 * Sample let Expression::
373 * Uninitialized let Variables::
375 The @code{if} Special Form
377 * if in more detail::
378 * type-of-animal in detail:: An example of an @code{if} expression.
380 Truth and Falsehood in Emacs Lisp
382 * nil explained:: @code{nil} has two meanings.
384 @code{save-excursion}
386 * Point and mark:: A review of various locations.
387 * Template for save-excursion::
389 A Few Buffer--Related Functions
391 * Finding More:: How to find more information.
392 * simplified-beginning-of-buffer:: Shows @code{goto-char},
393 @code{point-min}, and @code{push-mark}.
394 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
395 * append-to-buffer:: Uses @code{save-excursion} and
396 @code{insert-buffer-substring}.
397 * Buffer Related Review:: Review.
400 The Definition of @code{mark-whole-buffer}
402 * mark-whole-buffer overview::
403 * Body of mark-whole-buffer:: Only three lines of code.
405 The Definition of @code{append-to-buffer}
407 * append-to-buffer overview::
408 * append interactive:: A two part interactive expression.
409 * append-to-buffer body:: Incorporates a @code{let} expression.
410 * append save-excursion:: How the @code{save-excursion} works.
412 A Few More Complex Functions
414 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
415 * insert-buffer:: Read-only, and with @code{or}.
416 * beginning-of-buffer:: Shows @code{goto-char},
417 @code{point-min}, and @code{push-mark}.
418 * Second Buffer Related Review::
419 * optional Exercise::
421 The Definition of @code{insert-buffer}
423 * insert-buffer code::
424 * insert-buffer interactive:: When you can read, but not write.
425 * insert-buffer body:: The body has an @code{or} and a @code{let}.
426 * if & or:: Using an @code{if} instead of an @code{or}.
427 * Insert or:: How the @code{or} expression works.
428 * Insert let:: Two @code{save-excursion} expressions.
429 * New insert-buffer::
431 The Interactive Expression in @code{insert-buffer}
433 * Read-only buffer:: When a buffer cannot be modified.
434 * b for interactive:: An existing buffer or else its name.
436 Complete Definition of @code{beginning-of-buffer}
438 * Optional Arguments::
439 * beginning-of-buffer opt arg:: Example with optional argument.
440 * beginning-of-buffer complete::
442 @code{beginning-of-buffer} with an Argument
444 * Disentangle beginning-of-buffer::
445 * Large buffer case::
446 * Small buffer case::
448 Narrowing and Widening
450 * Narrowing advantages:: The advantages of narrowing
451 * save-restriction:: The @code{save-restriction} special form.
452 * what-line:: The number of the line that point is on.
455 @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
457 * Strange Names:: An historical aside: why the strange names?
458 * car & cdr:: Functions for extracting part of a list.
459 * cons:: Constructing a list.
460 * nthcdr:: Calling @code{cdr} repeatedly.
462 * setcar:: Changing the first element of a list.
463 * setcdr:: Changing the rest of a list.
469 * length:: How to find the length of a list.
471 Cutting and Storing Text
473 * Storing Text:: Text is stored in a list.
474 * zap-to-char:: Cutting out text up to a character.
475 * kill-region:: Cutting text out of a region.
476 * copy-region-as-kill:: A definition for copying text.
477 * Digression into C:: Minor note on C programming language macros.
478 * defvar:: How to give a variable an initial value.
479 * cons & search-fwd Review::
484 * Complete zap-to-char:: The complete implementation.
485 * zap-to-char interactive:: A three part interactive expression.
486 * zap-to-char body:: A short overview.
487 * search-forward:: How to search for a string.
488 * progn:: The @code{progn} special form.
489 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
493 * Complete kill-region:: The function definition.
494 * condition-case:: Dealing with a problem.
497 @code{copy-region-as-kill}
499 * Complete copy-region-as-kill:: The complete function definition.
500 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
502 The Body of @code{copy-region-as-kill}
504 * last-command & this-command::
505 * kill-append function::
506 * kill-new function::
508 Initializing a Variable with @code{defvar}
510 * See variable current value::
511 * defvar and asterisk::
513 How Lists are Implemented
516 * Symbols as Chest:: Exploring a powerful metaphor.
521 * Kill Ring Overview::
522 * kill-ring-yank-pointer:: The kill ring is a list.
523 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
527 * while:: Causing a stretch of code to repeat.
529 * Recursion:: Causing a function to call itself.
534 * Looping with while:: Repeat so long as test returns true.
535 * Loop Example:: A @code{while} loop that uses a list.
536 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
537 * Incrementing Loop:: A loop with an incrementing counter.
538 * Incrementing Loop Details::
539 * Decrementing Loop:: A loop with a decrementing counter.
541 Details of an Incrementing Loop
543 * Incrementing Example:: Counting pebbles in a triangle.
544 * Inc Example parts:: The parts of the function definition.
545 * Inc Example altogether:: Putting the function definition together.
547 Loop with a Decrementing Counter
549 * Decrementing Example:: More pebbles on the beach.
550 * Dec Example parts:: The parts of the function definition.
551 * Dec Example altogether:: Putting the function definition together.
553 Save your time: @code{dolist} and @code{dotimes}
560 * Building Robots:: Same model, different serial number ...
561 * Recursive Definition Parts:: Walk until you stop ...
562 * Recursion with list:: Using a list as the test whether to recurse.
563 * Recursive triangle function::
564 * Recursion with cond::
565 * Recursive Patterns:: Often used templates.
566 * No Deferment:: Don't store up work ...
567 * No deferment solution::
569 Recursion in Place of a Counter
571 * Recursive Example arg of 1 or 2::
572 * Recursive Example arg of 3 or 4::
580 Regular Expression Searches
582 * sentence-end:: The regular expression for @code{sentence-end}.
583 * re-search-forward:: Very similar to @code{search-forward}.
584 * forward-sentence:: A straightforward example of regexp search.
585 * forward-paragraph:: A somewhat complex example.
586 * etags:: How to create your own @file{TAGS} table.
588 * re-search Exercises::
590 @code{forward-sentence}
592 * Complete forward-sentence::
593 * fwd-sentence while loops:: Two @code{while} loops.
594 * fwd-sentence re-search:: A regular expression search.
596 @code{forward-paragraph}: a Goldmine of Functions
598 * forward-paragraph in brief:: Key parts of the function definition.
599 * fwd-para let:: The @code{let*} expression.
600 * fwd-para while:: The forward motion @code{while} loop.
602 Counting: Repetition and Regexps
605 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
606 * recursive-count-words:: Start with case of no words in region.
607 * Counting Exercise::
609 The @code{@value{COUNT-WORDS}} Function
611 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
612 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
614 Counting Words in a @code{defun}
616 * Divide and Conquer::
617 * Words and Symbols:: What to count?
618 * Syntax:: What constitutes a word or symbol?
619 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
620 * Several defuns:: Counting several defuns in a file.
621 * Find a File:: Do you want to look at a file?
622 * lengths-list-file:: A list of the lengths of many definitions.
623 * Several files:: Counting in definitions in different files.
624 * Several files recursively:: Recursively counting in different files.
625 * Prepare the data:: Prepare the data for display in a graph.
627 Count Words in @code{defuns} in Different Files
629 * lengths-list-many-files:: Return a list of the lengths of defuns.
630 * append:: Attach one list to another.
632 Prepare the Data for Display in a Graph
634 * Data for Display in Detail::
635 * Sorting:: Sorting lists.
636 * Files List:: Making a list of files.
637 * Counting function definitions::
641 * Columns of a graph::
642 * graph-body-print:: How to print the body of a graph.
643 * recursive-graph-body-print::
645 * Line Graph Exercise::
647 Your @file{.emacs} File
649 * Default Configuration::
650 * Site-wide Init:: You can write site-wide init files.
651 * defcustom:: Emacs will write code for you.
652 * Beginning init File:: How to write a @file{.emacs} init file.
653 * Text and Auto-fill:: Automatically wrap lines.
654 * Mail Aliases:: Use abbreviations for email addresses.
655 * Indent Tabs Mode:: Don't use tabs with @TeX{}
656 * Keybindings:: Create some personal keybindings.
657 * Keymaps:: More about key binding.
658 * Loading Files:: Load (i.e., evaluate) files automatically.
659 * Autoload:: Make functions available.
660 * Simple Extension:: Define a function; bind it to a key.
661 * X11 Colors:: Colors in X.
663 * Mode Line:: How to customize your mode line.
667 * debug:: How to use the built-in debugger.
668 * debug-on-entry:: Start debugging when you call a function.
669 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
670 * edebug:: How to use Edebug, a source level debugger.
671 * Debugging Exercises::
673 Handling the Kill Ring
675 * What the Kill Ring Does::
677 * yank:: Paste a copy of a clipped element.
678 * yank-pop:: Insert element pointed to.
681 The @code{current-kill} Function
683 * Code for current-kill::
684 * Understanding current-kill::
686 @code{current-kill} in Outline
688 * Body of current-kill::
689 * Digression concerning error:: How to mislead humans, but not computers.
690 * Determining the Element::
692 A Graph with Labeled Axes
695 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
696 * print-Y-axis:: Print a label for the vertical axis.
697 * print-X-axis:: Print a horizontal label.
698 * Print Whole Graph:: The function to print a complete graph.
700 The @code{print-Y-axis} Function
702 * print-Y-axis in Detail::
703 * Height of label:: What height for the Y axis?
704 * Compute a Remainder:: How to compute the remainder of a division.
705 * Y Axis Element:: Construct a line for the Y axis.
706 * Y-axis-column:: Generate a list of Y axis labels.
707 * print-Y-axis Penultimate:: A not quite final version.
709 The @code{print-X-axis} Function
711 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
712 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
714 Printing the Whole Graph
716 * The final version:: A few changes.
717 * Test print-graph:: Run a short test.
718 * Graphing words in defuns:: Executing the final code.
719 * lambda:: How to write an anonymous function.
720 * mapcar:: Apply a function to elements of a list.
721 * Another Bug:: Yet another bug @dots{} most insidious.
722 * Final printed graph:: The graph itself!
730 Most of the GNU Emacs integrated environment is written in the programming
731 language called Emacs Lisp. The code written in this programming
732 language is the software---the sets of instructions---that tell the
733 computer what to do when you give it commands. Emacs is designed so
734 that you can write new code in Emacs Lisp and easily install it as an
735 extension to the editor.
737 (GNU Emacs is sometimes called an ``extensible editor'', but it does
738 much more than provide editing capabilities. It is better to refer to
739 Emacs as an ``extensible computing environment''. However, that
740 phrase is quite a mouthful. It is easier to refer to Emacs simply as
741 an editor. Moreover, everything you do in Emacs---find the Mayan date
742 and phases of the moon, simplify polynomials, debug code, manage
743 files, read letters, write books---all these activities are kinds of
744 editing in the most general sense of the word.)
747 * Why:: Why learn Emacs Lisp?
748 * On Reading this Text:: Read, gain familiarity, pick up habits....
749 * Who You Are:: For whom this is written.
751 * Note for Novices:: You can read this as a novice.
757 @unnumberedsec Why Study Emacs Lisp?
760 Although Emacs Lisp is usually thought of in association only with Emacs,
761 it is a full computer programming language. You can use Emacs Lisp as
762 you would any other programming language.
764 Perhaps you want to understand programming; perhaps you want to extend
765 Emacs; or perhaps you want to become a programmer. This introduction to
766 Emacs Lisp is designed to get you started: to guide you in learning the
767 fundamentals of programming, and more importantly, to show you how you
768 can teach yourself to go further.
770 @node On Reading this Text
771 @unnumberedsec On Reading this Text
773 All through this document, you will see little sample programs you can
774 run inside of Emacs. If you read this document in Info inside of GNU
775 Emacs, you can run the programs as they appear. (This is easy to do and
776 is explained when the examples are presented.) Alternatively, you can
777 read this introduction as a printed book while sitting beside a computer
778 running Emacs. (This is what I like to do; I like printed books.) If
779 you don't have a running Emacs beside you, you can still read this book,
780 but in this case, it is best to treat it as a novel or as a travel guide
781 to a country not yet visited: interesting, but not the same as being
784 Much of this introduction is dedicated to walkthroughs or guided tours
785 of code used in GNU Emacs. These tours are designed for two purposes:
786 first, to give you familiarity with real, working code (code you use
787 every day); and, second, to give you familiarity with the way Emacs
788 works. It is interesting to see how a working environment is
791 hope that you will pick up the habit of browsing through source code.
792 You can learn from it and mine it for ideas. Having GNU Emacs is like
793 having a dragon's cave of treasures.
795 In addition to learning about Emacs as an editor and Emacs Lisp as a
796 programming language, the examples and guided tours will give you an
797 opportunity to get acquainted with Emacs as a Lisp programming
798 environment. GNU Emacs supports programming and provides tools that
799 you will want to become comfortable using, such as @kbd{M-.} (the key
800 which invokes the @code{find-tag} command). You will also learn about
801 buffers and other objects that are part of the environment.
802 Learning about these features of Emacs is like learning new routes
803 around your home town.
806 In addition, I have written several programs as extended examples.
807 Although these are examples, the programs are real. I use them.
808 Other people use them. You may use them. Beyond the fragments of
809 programs used for illustrations, there is very little in here that is
810 `just for teaching purposes'; what you see is used. This is a great
811 advantage of Emacs Lisp: it is easy to learn to use it for work.
814 Finally, I hope to convey some of the skills for using Emacs to
815 learn aspects of programming that you don't know. You can often use
816 Emacs to help you understand what puzzles you or to find out how to do
817 something new. This self-reliance is not only a pleasure, but an
821 @unnumberedsec For Whom This is Written
823 This text is written as an elementary introduction for people who are
824 not programmers. If you are a programmer, you may not be satisfied with
825 this primer. The reason is that you may have become expert at reading
826 reference manuals and be put off by the way this text is organized.
828 An expert programmer who reviewed this text said to me:
831 @i{I prefer to learn from reference manuals. I ``dive into'' each
832 paragraph, and ``come up for air'' between paragraphs.}
834 @i{When I get to the end of a paragraph, I assume that that subject is
835 done, finished, that I know everything I need (with the
836 possible exception of the case when the next paragraph starts talking
837 about it in more detail). I expect that a well written reference manual
838 will not have a lot of redundancy, and that it will have excellent
839 pointers to the (one) place where the information I want is.}
842 This introduction is not written for this person!
844 Firstly, I try to say everything at least three times: first, to
845 introduce it; second, to show it in context; and third, to show it in a
846 different context, or to review it.
848 Secondly, I hardly ever put all the information about a subject in one
849 place, much less in one paragraph. To my way of thinking, that imposes
850 too heavy a burden on the reader. Instead I try to explain only what
851 you need to know at the time. (Sometimes I include a little extra
852 information so you won't be surprised later when the additional
853 information is formally introduced.)
855 When you read this text, you are not expected to learn everything the
856 first time. Frequently, you need only make, as it were, a `nodding
857 acquaintance' with some of the items mentioned. My hope is that I have
858 structured the text and given you enough hints that you will be alert to
859 what is important, and concentrate on it.
861 You will need to ``dive into'' some paragraphs; there is no other way
862 to read them. But I have tried to keep down the number of such
863 paragraphs. This book is intended as an approachable hill, rather than
864 as a daunting mountain.
866 This introduction to @cite{Programming in Emacs Lisp} has a companion
869 @cite{The GNU Emacs Lisp Reference Manual}.
872 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
873 Emacs Lisp Reference Manual}.
875 The reference manual has more detail than this introduction. In the
876 reference manual, all the information about one topic is concentrated
877 in one place. You should turn to it if you are like the programmer
878 quoted above. And, of course, after you have read this
879 @cite{Introduction}, you will find the @cite{Reference Manual} useful
880 when you are writing your own programs.
883 @unnumberedsec Lisp History
886 Lisp was first developed in the late 1950s at the Massachusetts
887 Institute of Technology for research in artificial intelligence. The
888 great power of the Lisp language makes it superior for other purposes as
889 well, such as writing editor commands and integrated environments.
893 GNU Emacs Lisp is largely inspired by Maclisp, which was written at MIT
894 in the 1960s. It is somewhat inspired by Common Lisp, which became a
895 standard in the 1980s. However, Emacs Lisp is much simpler than Common
896 Lisp. (The standard Emacs distribution contains an optional extensions
897 file, @file{cl.el}, that adds many Common Lisp features to Emacs Lisp.)
899 @node Note for Novices
900 @unnumberedsec A Note for Novices
902 If you don't know GNU Emacs, you can still read this document
903 profitably. However, I recommend you learn Emacs, if only to learn to
904 move around your computer screen. You can teach yourself how to use
905 Emacs with the on-line tutorial. To use it, type @kbd{C-h t}. (This
906 means you press and release the @key{CTRL} key and the @kbd{h} at the
907 same time, and then press and release @kbd{t}.)
909 Also, I often refer to one of Emacs's standard commands by listing the
910 keys which you press to invoke the command and then giving the name of
911 the command in parentheses, like this: @kbd{M-C-\}
912 (@code{indent-region}). What this means is that the
913 @code{indent-region} command is customarily invoked by typing
914 @kbd{M-C-\}. (You can, if you wish, change the keys that are typed to
915 invoke the command; this is called @dfn{rebinding}. @xref{Keymaps, ,
916 Keymaps}.) The abbreviation @kbd{M-C-\} means that you type your
917 @key{META} key, @key{CTRL} key and @key{\} key all at the same time.
918 (On many modern keyboards the @key{META} key is labeled
920 Sometimes a combination like this is called a keychord, since it is
921 similar to the way you play a chord on a piano. If your keyboard does
922 not have a @key{META} key, the @key{ESC} key prefix is used in place
923 of it. In this case, @kbd{M-C-\} means that you press and release your
924 @key{ESC} key and then type the @key{CTRL} key and the @key{\} key at
925 the same time. But usually @kbd{M-C-\} means press the @key{CTRL} key
926 along with the key that is labeled @key{ALT} and, at the same time,
927 press the @key{\} key.
929 In addition to typing a lone keychord, you can prefix what you type
930 with @kbd{C-u}, which is called the `universal argument'. The
931 @kbd{C-u} keychord passes an argument to the subsequent command.
932 Thus, to indent a region of plain text by 6 spaces, mark the region,
933 and then type @w{@kbd{C-u 6 M-C-\}}. (If you do not specify a number,
934 Emacs either passes the number 4 to the command or otherwise runs the
935 command differently than it would otherwise.) @xref{Arguments, ,
936 Numeric Arguments, emacs, The GNU Emacs Manual}.
938 If you are reading this in Info using GNU Emacs, you can read through
939 this whole document just by pressing the space bar, @key{SPC}.
940 (To learn about Info, type @kbd{C-h i} and then select Info.)
942 A note on terminology: when I use the word Lisp alone, I often am
943 referring to the various dialects of Lisp in general, but when I speak
944 of Emacs Lisp, I am referring to GNU Emacs Lisp in particular.
947 @unnumberedsec Thank You
949 My thanks to all who helped me with this book. My especial thanks to
950 @r{Jim Blandy}, @r{Noah Friedman}, @w{Jim Kingdon}, @r{Roland
951 McGrath}, @w{Frank Ritter}, @w{Randy Smith}, @w{Richard M.
952 Stallman}, and @w{Melissa Weisshaus}. My thanks also go to both
953 @w{Philip Johnson} and @w{David Stampe} for their patient
954 encouragement. My mistakes are my own.
966 @c ================ Beginning of main text ================
968 @c Start main text on right-hand (verso) page
971 \par\vfill\supereject
974 \par\vfill\supereject
976 \par\vfill\supereject
978 \par\vfill\supereject
982 @c Note: this resetting of the page number back to 1 causes TeX to gripe
983 @c about already having seen page numbers 1-4 before (in the preface):
984 @c pdfTeX warning (ext4): destination with the same identifier (name{1})
985 @c has been already used, duplicate ignored
986 @c I guess that is harmless (what happens if a later part of the text
987 @c makes a link to something in the first 4 pages though?).
988 @c E.g., note that the Emacs manual has a preface, but does not bother
989 @c resetting the page numbers back to 1 after that.
992 @evenheading @thispage @| @| @thischapter
993 @oddheading @thissection @| @| @thispage
997 @node List Processing
998 @chapter List Processing
1000 To the untutored eye, Lisp is a strange programming language. In Lisp
1001 code there are parentheses everywhere. Some people even claim that
1002 the name stands for `Lots of Isolated Silly Parentheses'. But the
1003 claim is unwarranted. Lisp stands for LISt Processing, and the
1004 programming language handles @emph{lists} (and lists of lists) by
1005 putting them between parentheses. The parentheses mark the boundaries
1006 of the list. Sometimes a list is preceded by a single apostrophe or
1007 quotation mark, @samp{'}@footnote{The single apostrophe or quotation
1008 mark is an abbreviation for the function @code{quote}; you need not
1009 think about functions now; functions are defined in @ref{Making
1010 Errors, , Generate an Error Message}.} Lists are the basis of Lisp.
1013 * Lisp Lists:: What are lists?
1014 * Run a Program:: Any list in Lisp is a program ready to run.
1015 * Making Errors:: Generating an error message.
1016 * Names & Definitions:: Names of symbols and function definitions.
1017 * Lisp Interpreter:: What the Lisp interpreter does.
1018 * Evaluation:: Running a program.
1019 * Variables:: Returning a value from a variable.
1020 * Arguments:: Passing information to a function.
1021 * set & setq:: Setting the value of a variable.
1022 * Summary:: The major points.
1023 * Error Message Exercises::
1030 In Lisp, a list looks like this: @code{'(rose violet daisy buttercup)}.
1031 This list is preceded by a single apostrophe. It could just as well be
1032 written as follows, which looks more like the kind of list you are likely
1033 to be familiar with:
1045 The elements of this list are the names of the four different flowers,
1046 separated from each other by whitespace and surrounded by parentheses,
1047 like flowers in a field with a stone wall around them.
1048 @cindex Flowers in a field
1051 * Numbers Lists:: List have numbers, other lists, in them.
1052 * Lisp Atoms:: Elemental entities.
1053 * Whitespace in Lists:: Formatting lists to be readable.
1054 * Typing Lists:: How GNU Emacs helps you type lists.
1059 @unnumberedsubsec Numbers, Lists inside of Lists
1062 Lists can also have numbers in them, as in this list: @code{(+ 2 2)}.
1063 This list has a plus-sign, @samp{+}, followed by two @samp{2}s, each
1064 separated by whitespace.
1066 In Lisp, both data and programs are represented the same way; that is,
1067 they are both lists of words, numbers, or other lists, separated by
1068 whitespace and surrounded by parentheses. (Since a program looks like
1069 data, one program may easily serve as data for another; this is a very
1070 powerful feature of Lisp.) (Incidentally, these two parenthetical
1071 remarks are @emph{not} Lisp lists, because they contain @samp{;} and
1072 @samp{.} as punctuation marks.)
1075 Here is another list, this time with a list inside of it:
1078 '(this list has (a list inside of it))
1081 The components of this list are the words @samp{this}, @samp{list},
1082 @samp{has}, and the list @samp{(a list inside of it)}. The interior
1083 list is made up of the words @samp{a}, @samp{list}, @samp{inside},
1084 @samp{of}, @samp{it}.
1087 @subsection Lisp Atoms
1090 In Lisp, what we have been calling words are called @dfn{atoms}. This
1091 term comes from the historical meaning of the word atom, which means
1092 `indivisible'. As far as Lisp is concerned, the words we have been
1093 using in the lists cannot be divided into any smaller parts and still
1094 mean the same thing as part of a program; likewise with numbers and
1095 single character symbols like @samp{+}. On the other hand, unlike an
1096 ancient atom, a list can be split into parts. (@xref{car cdr & cons,
1097 , @code{car} @code{cdr} & @code{cons} Fundamental Functions}.)
1099 In a list, atoms are separated from each other by whitespace. They can be
1100 right next to a parenthesis.
1102 @cindex @samp{empty list} defined
1103 Technically speaking, a list in Lisp consists of parentheses surrounding
1104 atoms separated by whitespace or surrounding other lists or surrounding
1105 both atoms and other lists. A list can have just one atom in it or
1106 have nothing in it at all. A list with nothing in it looks like this:
1107 @code{()}, and is called the @dfn{empty list}. Unlike anything else, an
1108 empty list is considered both an atom and a list at the same time.
1110 @cindex Symbolic expressions, introduced
1111 @cindex @samp{expression} defined
1112 @cindex @samp{form} defined
1113 The printed representation of both atoms and lists are called
1114 @dfn{symbolic expressions} or, more concisely, @dfn{s-expressions}.
1115 The word @dfn{expression} by itself can refer to either the printed
1116 representation, or to the atom or list as it is held internally in the
1117 computer. Often, people use the term @dfn{expression}
1118 indiscriminately. (Also, in many texts, the word @dfn{form} is used
1119 as a synonym for expression.)
1121 Incidentally, the atoms that make up our universe were named such when
1122 they were thought to be indivisible; but it has been found that physical
1123 atoms are not indivisible. Parts can split off an atom or it can
1124 fission into two parts of roughly equal size. Physical atoms were named
1125 prematurely, before their truer nature was found. In Lisp, certain
1126 kinds of atom, such as an array, can be separated into parts; but the
1127 mechanism for doing this is different from the mechanism for splitting a
1128 list. As far as list operations are concerned, the atoms of a list are
1131 As in English, the meanings of the component letters of a Lisp atom
1132 are different from the meaning the letters make as a word. For
1133 example, the word for the South American sloth, the @samp{ai}, is
1134 completely different from the two words, @samp{a}, and @samp{i}.
1136 There are many kinds of atom in nature but only a few in Lisp: for
1137 example, @dfn{numbers}, such as 37, 511, or 1729, and @dfn{symbols}, such
1138 as @samp{+}, @samp{foo}, or @samp{forward-line}. The words we have
1139 listed in the examples above are all symbols. In everyday Lisp
1140 conversation, the word ``atom'' is not often used, because programmers
1141 usually try to be more specific about what kind of atom they are dealing
1142 with. Lisp programming is mostly about symbols (and sometimes numbers)
1143 within lists. (Incidentally, the preceding three word parenthetical
1144 remark is a proper list in Lisp, since it consists of atoms, which in
1145 this case are symbols, separated by whitespace and enclosed by
1146 parentheses, without any non-Lisp punctuation.)
1149 Text between double quotation marks---even sentences or
1150 paragraphs---is also an atom. Here is an example:
1151 @cindex Text between double quotation marks
1154 '(this list includes "text between quotation marks.")
1157 @cindex @samp{string} defined
1159 In Lisp, all of the quoted text including the punctuation mark and the
1160 blank spaces is a single atom. This kind of atom is called a
1161 @dfn{string} (for `string of characters') and is the sort of thing that
1162 is used for messages that a computer can print for a human to read.
1163 Strings are a different kind of atom than numbers or symbols and are
1166 @node Whitespace in Lists
1167 @subsection Whitespace in Lists
1168 @cindex Whitespace in lists
1171 The amount of whitespace in a list does not matter. From the point of view
1172 of the Lisp language,
1183 is exactly the same as this:
1186 '(this list looks like this)
1189 Both examples show what to Lisp is the same list, the list made up of
1190 the symbols @samp{this}, @samp{list}, @samp{looks}, @samp{like}, and
1191 @samp{this} in that order.
1193 Extra whitespace and newlines are designed to make a list more readable
1194 by humans. When Lisp reads the expression, it gets rid of all the extra
1195 whitespace (but it needs to have at least one space between atoms in
1196 order to tell them apart.)
1198 Odd as it seems, the examples we have seen cover almost all of what Lisp
1199 lists look like! Every other list in Lisp looks more or less like one
1200 of these examples, except that the list may be longer and more complex.
1201 In brief, a list is between parentheses, a string is between quotation
1202 marks, a symbol looks like a word, and a number looks like a number.
1203 (For certain situations, square brackets, dots and a few other special
1204 characters may be used; however, we will go quite far without them.)
1207 @subsection GNU Emacs Helps You Type Lists
1208 @cindex Help typing lists
1209 @cindex Formatting help
1211 When you type a Lisp expression in GNU Emacs using either Lisp
1212 Interaction mode or Emacs Lisp mode, you have available to you several
1213 commands to format the Lisp expression so it is easy to read. For
1214 example, pressing the @key{TAB} key automatically indents the line the
1215 cursor is on by the right amount. A command to properly indent the
1216 code in a region is customarily bound to @kbd{M-C-\}. Indentation is
1217 designed so that you can see which elements of a list belong to which
1218 list---elements of a sub-list are indented more than the elements of
1221 In addition, when you type a closing parenthesis, Emacs momentarily
1222 jumps the cursor back to the matching opening parenthesis, so you can
1223 see which one it is. This is very useful, since every list you type
1224 in Lisp must have its closing parenthesis match its opening
1225 parenthesis. (@xref{Major Modes, , Major Modes, emacs, The GNU Emacs
1226 Manual}, for more information about Emacs's modes.)
1229 @section Run a Program
1230 @cindex Run a program
1231 @cindex Program, running one
1233 @cindex @samp{evaluate} defined
1234 A list in Lisp---any list---is a program ready to run. If you run it
1235 (for which the Lisp jargon is @dfn{evaluate}), the computer will do one
1236 of three things: do nothing except return to you the list itself; send
1237 you an error message; or, treat the first symbol in the list as a
1238 command to do something. (Usually, of course, it is the last of these
1239 three things that you really want!)
1241 @c use code for the single apostrophe, not samp.
1242 The single apostrophe, @code{'}, that I put in front of some of the
1243 example lists in preceding sections is called a @dfn{quote}; when it
1244 precedes a list, it tells Lisp to do nothing with the list, other than
1245 take it as it is written. But if there is no quote preceding a list,
1246 the first item of the list is special: it is a command for the computer
1247 to obey. (In Lisp, these commands are called @emph{functions}.) The list
1248 @code{(+ 2 2)} shown above did not have a quote in front of it, so Lisp
1249 understands that the @code{+} is an instruction to do something with the
1250 rest of the list: add the numbers that follow.
1253 If you are reading this inside of GNU Emacs in Info, here is how you can
1254 evaluate such a list: place your cursor immediately after the right
1255 hand parenthesis of the following list and then type @kbd{C-x C-e}:
1261 @c use code for the number four, not samp.
1263 You will see the number @code{4} appear in the echo area. (In the
1264 jargon, what you have just done is ``evaluate the list.'' The echo area
1265 is the line at the bottom of the screen that displays or ``echoes''
1266 text.) Now try the same thing with a quoted list: place the cursor
1267 right after the following list and type @kbd{C-x C-e}:
1270 '(this is a quoted list)
1274 You will see @code{(this is a quoted list)} appear in the echo area.
1276 @cindex Lisp interpreter, explained
1277 @cindex Interpreter, Lisp, explained
1278 In both cases, what you are doing is giving a command to the program
1279 inside of GNU Emacs called the @dfn{Lisp interpreter}---giving the
1280 interpreter a command to evaluate the expression. The name of the Lisp
1281 interpreter comes from the word for the task done by a human who comes
1282 up with the meaning of an expression---who ``interprets'' it.
1284 You can also evaluate an atom that is not part of a list---one that is
1285 not surrounded by parentheses; again, the Lisp interpreter translates
1286 from the humanly readable expression to the language of the computer.
1287 But before discussing this (@pxref{Variables}), we will discuss what the
1288 Lisp interpreter does when you make an error.
1291 @section Generate an Error Message
1292 @cindex Generate an error message
1293 @cindex Error message generation
1295 Partly so you won't worry if you do it accidentally, we will now give
1296 a command to the Lisp interpreter that generates an error message.
1297 This is a harmless activity; and indeed, we will often try to generate
1298 error messages intentionally. Once you understand the jargon, error
1299 messages can be informative. Instead of being called ``error''
1300 messages, they should be called ``help'' messages. They are like
1301 signposts to a traveler in a strange country; deciphering them can be
1302 hard, but once understood, they can point the way.
1304 The error message is generated by a built-in GNU Emacs debugger. We
1305 will `enter the debugger'. You get out of the debugger by typing @code{q}.
1307 What we will do is evaluate a list that is not quoted and does not
1308 have a meaningful command as its first element. Here is a list almost
1309 exactly the same as the one we just used, but without the single-quote
1310 in front of it. Position the cursor right after it and type @kbd{C-x
1314 (this is an unquoted list)
1319 What you see depends on which version of Emacs you are running. GNU
1320 Emacs version 22 provides more information than version 20 and before.
1321 First, the more recent result of generating an error; then the
1322 earlier, version 20 result.
1326 In GNU Emacs version 22, a @file{*Backtrace*} window will open up and
1327 you will see the following in it:
1330 A @file{*Backtrace*} window will open up and you should see the
1335 ---------- Buffer: *Backtrace* ----------
1336 Debugger entered--Lisp error: (void-function this)
1337 (this is an unquoted list)
1338 eval((this is an unquoted list))
1339 eval-last-sexp-1(nil)
1341 call-interactively(eval-last-sexp)
1342 ---------- Buffer: *Backtrace* ----------
1348 Your cursor will be in this window (you may have to wait a few seconds
1349 before it becomes visible). To quit the debugger and make the
1350 debugger window go away, type:
1357 Please type @kbd{q} right now, so you become confident that you can
1358 get out of the debugger. Then, type @kbd{C-x C-e} again to re-enter
1361 @cindex @samp{function} defined
1362 Based on what we already know, we can almost read this error message.
1364 You read the @file{*Backtrace*} buffer from the bottom up; it tells
1365 you what Emacs did. When you typed @kbd{C-x C-e}, you made an
1366 interactive call to the command @code{eval-last-sexp}. @code{eval} is
1367 an abbreviation for `evaluate' and @code{sexp} is an abbreviation for
1368 `symbolic expression'. The command means `evaluate last symbolic
1369 expression', which is the expression just before your cursor.
1371 Each line above tells you what the Lisp interpreter evaluated next.
1372 The most recent action is at the top. The buffer is called the
1373 @file{*Backtrace*} buffer because it enables you to track Emacs
1377 At the top of the @file{*Backtrace*} buffer, you see the line:
1380 Debugger entered--Lisp error: (void-function this)
1384 The Lisp interpreter tried to evaluate the first atom of the list, the
1385 word @samp{this}. It is this action that generated the error message
1386 @samp{void-function this}.
1388 The message contains the words @samp{void-function} and @samp{this}.
1390 @cindex @samp{function} defined
1391 The word @samp{function} was mentioned once before. It is a very
1392 important word. For our purposes, we can define it by saying that a
1393 @dfn{function} is a set of instructions to the computer that tell the
1394 computer to do something.
1396 Now we can begin to understand the error message: @samp{void-function
1397 this}. The function (that is, the word @samp{this}) does not have a
1398 definition of any set of instructions for the computer to carry out.
1400 The slightly odd word, @samp{void-function}, is designed to cover the
1401 way Emacs Lisp is implemented, which is that when a symbol does not
1402 have a function definition attached to it, the place that should
1403 contain the instructions is `void'.
1405 On the other hand, since we were able to add 2 plus 2 successfully, by
1406 evaluating @code{(+ 2 2)}, we can infer that the symbol @code{+} must
1407 have a set of instructions for the computer to obey and those
1408 instructions must be to add the numbers that follow the @code{+}.
1410 It is possible to prevent Emacs entering the debugger in cases like
1411 this. We do not explain how to do that here, but we will mention what
1412 the result looks like, because you may encounter a similar situation
1413 if there is a bug in some Emacs code that you are using. In such
1414 cases, you will see only one line of error message; it will appear in
1415 the echo area and look like this:
1418 Symbol's function definition is void:@: this
1423 (Also, your terminal may beep at you---some do, some don't; and others
1424 blink. This is just a device to get your attention.)
1426 The message goes away as soon as you type a key, even just to
1429 We know the meaning of the word @samp{Symbol}. It refers to the first
1430 atom of the list, the word @samp{this}. The word @samp{function}
1431 refers to the instructions that tell the computer what to do.
1432 (Technically, the symbol tells the computer where to find the
1433 instructions, but this is a complication we can ignore for the
1436 The error message can be understood: @samp{Symbol's function
1437 definition is void:@: this}. The symbol (that is, the word
1438 @samp{this}) lacks instructions for the computer to carry out.
1440 @node Names & Definitions
1441 @section Symbol Names and Function Definitions
1442 @cindex Symbol names
1444 We can articulate another characteristic of Lisp based on what we have
1445 discussed so far---an important characteristic: a symbol, like
1446 @code{+}, is not itself the set of instructions for the computer to
1447 carry out. Instead, the symbol is used, perhaps temporarily, as a way
1448 of locating the definition or set of instructions. What we see is the
1449 name through which the instructions can be found. Names of people
1450 work the same way. I can be referred to as @samp{Bob}; however, I am
1451 not the letters @samp{B}, @samp{o}, @samp{b} but am, or was, the
1452 consciousness consistently associated with a particular life-form.
1453 The name is not me, but it can be used to refer to me.
1455 In Lisp, one set of instructions can be attached to several names.
1456 For example, the computer instructions for adding numbers can be
1457 linked to the symbol @code{plus} as well as to the symbol @code{+}
1458 (and are in some dialects of Lisp). Among humans, I can be referred
1459 to as @samp{Robert} as well as @samp{Bob} and by other words as well.
1461 On the other hand, a symbol can have only one function definition
1462 attached to it at a time. Otherwise, the computer would be confused as
1463 to which definition to use. If this were the case among people, only
1464 one person in the world could be named @samp{Bob}. However, the function
1465 definition to which the name refers can be changed readily.
1466 (@xref{Install, , Install a Function Definition}.)
1468 Since Emacs Lisp is large, it is customary to name symbols in a way
1469 that identifies the part of Emacs to which the function belongs.
1470 Thus, all the names for functions that deal with Texinfo start with
1471 @samp{texinfo-} and those for functions that deal with reading mail
1472 start with @samp{rmail-}.
1474 @node Lisp Interpreter
1475 @section The Lisp Interpreter
1476 @cindex Lisp interpreter, what it does
1477 @cindex Interpreter, what it does
1479 Based on what we have seen, we can now start to figure out what the
1480 Lisp interpreter does when we command it to evaluate a list.
1481 First, it looks to see whether there is a quote before the list; if
1482 there is, the interpreter just gives us the list. On the other
1483 hand, if there is no quote, the interpreter looks at the first element
1484 in the list and sees whether it has a function definition. If it does,
1485 the interpreter carries out the instructions in the function definition.
1486 Otherwise, the interpreter prints an error message.
1488 This is how Lisp works. Simple. There are added complications which we
1489 will get to in a minute, but these are the fundamentals. Of course, to
1490 write Lisp programs, you need to know how to write function definitions
1491 and attach them to names, and how to do this without confusing either
1492 yourself or the computer.
1495 * Complications:: Variables, Special forms, Lists within.
1496 * Byte Compiling:: Specially processing code for speed.
1501 @unnumberedsubsec Complications
1504 Now, for the first complication. In addition to lists, the Lisp
1505 interpreter can evaluate a symbol that is not quoted and does not have
1506 parentheses around it. The Lisp interpreter will attempt to determine
1507 the symbol's value as a @dfn{variable}. This situation is described
1508 in the section on variables. (@xref{Variables}.)
1510 @cindex Special form
1511 The second complication occurs because some functions are unusual and
1512 do not work in the usual manner. Those that don't are called
1513 @dfn{special forms}. They are used for special jobs, like defining a
1514 function, and there are not many of them. In the next few chapters,
1515 you will be introduced to several of the more important special forms.
1517 As well as special forms, there are also @dfn{macros}. A macro
1518 is a construct defined in Lisp, which differs from a function in that it
1519 translates a Lisp expression into another expression that is to be
1520 evaluated in place of the original expression. (@xref{Lisp macro}.)
1522 For the purposes of this introduction, you do not need to worry too much
1523 about whether something is a special form, macro, or ordinary function.
1524 For example, @code{if} is a special form (@pxref{if}), but @code{when}
1525 is a macro (@pxref{Lisp macro}). In earlier versions of Emacs,
1526 @code{defun} was a special form, but now it is a macro (@pxref{defun}).
1527 It still behaves in the same way.
1529 The final complication is this: if the function that the
1530 Lisp interpreter is looking at is not a special form, and if it is part
1531 of a list, the Lisp interpreter looks to see whether the list has a list
1532 inside of it. If there is an inner list, the Lisp interpreter first
1533 figures out what it should do with the inside list, and then it works on
1534 the outside list. If there is yet another list embedded inside the
1535 inner list, it works on that one first, and so on. It always works on
1536 the innermost list first. The interpreter works on the innermost list
1537 first, to evaluate the result of that list. The result may be
1538 used by the enclosing expression.
1540 Otherwise, the interpreter works left to right, from one expression to
1543 @node Byte Compiling
1544 @subsection Byte Compiling
1545 @cindex Byte compiling
1547 One other aspect of interpreting: the Lisp interpreter is able to
1548 interpret two kinds of entity: humanly readable code, on which we will
1549 focus exclusively, and specially processed code, called @dfn{byte
1550 compiled} code, which is not humanly readable. Byte compiled code
1551 runs faster than humanly readable code.
1553 You can transform humanly readable code into byte compiled code by
1554 running one of the compile commands such as @code{byte-compile-file}.
1555 Byte compiled code is usually stored in a file that ends with a
1556 @file{.elc} extension rather than a @file{.el} extension. You will
1557 see both kinds of file in the @file{emacs/lisp} directory; the files
1558 to read are those with @file{.el} extensions.
1560 As a practical matter, for most things you might do to customize or
1561 extend Emacs, you do not need to byte compile; and I will not discuss
1562 the topic here. @xref{Byte Compilation, , Byte Compilation, elisp,
1563 The GNU Emacs Lisp Reference Manual}, for a full description of byte
1570 When the Lisp interpreter works on an expression, the term for the
1571 activity is called @dfn{evaluation}. We say that the interpreter
1572 `evaluates the expression'. I've used this term several times before.
1573 The word comes from its use in everyday language, `to ascertain the
1574 value or amount of; to appraise', according to @cite{Webster's New
1575 Collegiate Dictionary}.
1578 * How the Interpreter Acts:: Returns and Side Effects...
1579 * Evaluating Inner Lists:: Lists within lists...
1583 @node How the Interpreter Acts
1584 @unnumberedsubsec How the Lisp Interpreter Acts
1587 @cindex @samp{returned value} explained
1588 After evaluating an expression, the Lisp interpreter will most likely
1589 @dfn{return} the value that the computer produces by carrying out the
1590 instructions it found in the function definition, or perhaps it will
1591 give up on that function and produce an error message. (The interpreter
1592 may also find itself tossed, so to speak, to a different function or it
1593 may attempt to repeat continually what it is doing for ever and ever in
1594 what is called an `infinite loop'. These actions are less common; and
1595 we can ignore them.) Most frequently, the interpreter returns a value.
1597 @cindex @samp{side effect} defined
1598 At the same time the interpreter returns a value, it may do something
1599 else as well, such as move a cursor or copy a file; this other kind of
1600 action is called a @dfn{side effect}. Actions that we humans think are
1601 important, such as printing results, are often ``side effects'' to the
1602 Lisp interpreter. The jargon can sound peculiar, but it turns out that
1603 it is fairly easy to learn to use side effects.
1605 In summary, evaluating a symbolic expression most commonly causes the
1606 Lisp interpreter to return a value and perhaps carry out a side effect;
1607 or else produce an error.
1609 @node Evaluating Inner Lists
1610 @subsection Evaluating Inner Lists
1611 @cindex Inner list evaluation
1612 @cindex Evaluating inner lists
1614 If evaluation applies to a list that is inside another list, the outer
1615 list may use the value returned by the first evaluation as information
1616 when the outer list is evaluated. This explains why inner expressions
1617 are evaluated first: the values they return are used by the outer
1621 We can investigate this process by evaluating another addition example.
1622 Place your cursor after the following expression and type @kbd{C-x C-e}:
1629 The number 8 will appear in the echo area.
1631 What happens is that the Lisp interpreter first evaluates the inner
1632 expression, @code{(+ 3 3)}, for which the value 6 is returned; then it
1633 evaluates the outer expression as if it were written @code{(+ 2 6)}, which
1634 returns the value 8. Since there are no more enclosing expressions to
1635 evaluate, the interpreter prints that value in the echo area.
1637 Now it is easy to understand the name of the command invoked by the
1638 keystrokes @kbd{C-x C-e}: the name is @code{eval-last-sexp}. The
1639 letters @code{sexp} are an abbreviation for `symbolic expression', and
1640 @code{eval} is an abbreviation for `evaluate'. The command means
1641 `evaluate last symbolic expression'.
1643 As an experiment, you can try evaluating the expression by putting the
1644 cursor at the beginning of the next line immediately following the
1645 expression, or inside the expression.
1648 Here is another copy of the expression:
1655 If you place the cursor at the beginning of the blank line that
1656 immediately follows the expression and type @kbd{C-x C-e}, you will
1657 still get the value 8 printed in the echo area. Now try putting the
1658 cursor inside the expression. If you put it right after the next to
1659 last parenthesis (so it appears to sit on top of the last parenthesis),
1660 you will get a 6 printed in the echo area! This is because the command
1661 evaluates the expression @code{(+ 3 3)}.
1663 Now put the cursor immediately after a number. Type @kbd{C-x C-e} and
1664 you will get the number itself. In Lisp, if you evaluate a number, you
1665 get the number itself---this is how numbers differ from symbols. If you
1666 evaluate a list starting with a symbol like @code{+}, you will get a
1667 value returned that is the result of the computer carrying out the
1668 instructions in the function definition attached to that name. If a
1669 symbol by itself is evaluated, something different happens, as we will
1670 see in the next section.
1676 In Emacs Lisp, a symbol can have a value attached to it just as it can
1677 have a function definition attached to it. The two are different.
1678 The function definition is a set of instructions that a computer will
1679 obey. A value, on the other hand, is something, such as number or a
1680 name, that can vary (which is why such a symbol is called a variable).
1681 The value of a symbol can be any expression in Lisp, such as a symbol,
1682 number, list, or string. A symbol that has a value is often called a
1685 A symbol can have both a function definition and a value attached to
1686 it at the same time. Or it can have just one or the other.
1687 The two are separate. This is somewhat similar
1688 to the way the name Cambridge can refer to the city in Massachusetts
1689 and have some information attached to the name as well, such as
1690 ``great programming center''.
1693 (Incidentally, in Emacs Lisp, a symbol can have two
1694 other things attached to it, too: a property list and a documentation
1695 string; these are discussed later.)
1698 Another way to think about this is to imagine a symbol as being a chest
1699 of drawers. The function definition is put in one drawer, the value in
1700 another, and so on. What is put in the drawer holding the value can be
1701 changed without affecting the contents of the drawer holding the
1702 function definition, and vice-verse.
1705 * fill-column Example::
1706 * Void Function:: The error message for a symbol
1708 * Void Variable:: The error message for a symbol without a value.
1712 @node fill-column Example
1713 @unnumberedsubsec @code{fill-column}, an Example Variable
1716 @findex fill-column, @r{an example variable}
1717 @cindex Example variable, @code{fill-column}
1718 @cindex Variable, example of, @code{fill-column}
1719 The variable @code{fill-column} illustrates a symbol with a value
1720 attached to it: in every GNU Emacs buffer, this symbol is set to some
1721 value, usually 72 or 70, but sometimes to some other value. To find the
1722 value of this symbol, evaluate it by itself. If you are reading this in
1723 Info inside of GNU Emacs, you can do this by putting the cursor after
1724 the symbol and typing @kbd{C-x C-e}:
1731 After I typed @kbd{C-x C-e}, Emacs printed the number 72 in my echo
1732 area. This is the value for which @code{fill-column} is set for me as I
1733 write this. It may be different for you in your Info buffer. Notice
1734 that the value returned as a variable is printed in exactly the same way
1735 as the value returned by a function carrying out its instructions. From
1736 the point of view of the Lisp interpreter, a value returned is a value
1737 returned. What kind of expression it came from ceases to matter once
1740 A symbol can have any value attached to it or, to use the jargon, we can
1741 @dfn{bind} the variable to a value: to a number, such as 72; to a
1742 string, @code{"such as this"}; to a list, such as @code{(spruce pine
1743 oak)}; we can even bind a variable to a function definition.
1745 A symbol can be bound to a value in several ways. @xref{set & setq, ,
1746 Setting the Value of a Variable}, for information about one way to do
1750 @subsection Error Message for a Symbol Without a Function
1751 @cindex Symbol without function error
1752 @cindex Error for symbol without function
1754 When we evaluated @code{fill-column} to find its value as a variable,
1755 we did not place parentheses around the word. This is because we did
1756 not intend to use it as a function name.
1758 If @code{fill-column} were the first or only element of a list, the
1759 Lisp interpreter would attempt to find the function definition
1760 attached to it. But @code{fill-column} has no function definition.
1761 Try evaluating this:
1769 You will create a @file{*Backtrace*} buffer that says:
1773 ---------- Buffer: *Backtrace* ----------
1774 Debugger entered--Lisp error: (void-function fill-column)
1777 eval-last-sexp-1(nil)
1779 call-interactively(eval-last-sexp)
1780 ---------- Buffer: *Backtrace* ----------
1785 (Remember, to quit the debugger and make the debugger window go away,
1786 type @kbd{q} in the @file{*Backtrace*} buffer.)
1790 In GNU Emacs 20 and before, you will produce an error message that says:
1793 Symbol's function definition is void:@: fill-column
1797 (The message will go away as soon as you move the cursor or type
1802 @subsection Error Message for a Symbol Without a Value
1803 @cindex Symbol without value error
1804 @cindex Error for symbol without value
1806 If you attempt to evaluate a symbol that does not have a value bound to
1807 it, you will receive an error message. You can see this by
1808 experimenting with our 2 plus 2 addition. In the following expression,
1809 put your cursor right after the @code{+}, before the first number 2,
1818 In GNU Emacs 22, you will create a @file{*Backtrace*} buffer that
1823 ---------- Buffer: *Backtrace* ----------
1824 Debugger entered--Lisp error: (void-variable +)
1826 eval-last-sexp-1(nil)
1828 call-interactively(eval-last-sexp)
1829 ---------- Buffer: *Backtrace* ----------
1834 (Again, you can quit the debugger by
1835 typing @kbd{q} in the @file{*Backtrace*} buffer.)
1837 This backtrace is different from the very first error message we saw,
1838 which said, @samp{Debugger entered--Lisp error: (void-function this)}.
1839 In this case, the function does not have a value as a variable; while
1840 in the other error message, the function (the word `this') did not
1843 In this experiment with the @code{+}, what we did was cause the Lisp
1844 interpreter to evaluate the @code{+} and look for the value of the
1845 variable instead of the function definition. We did this by placing the
1846 cursor right after the symbol rather than after the parenthesis of the
1847 enclosing list as we did before. As a consequence, the Lisp interpreter
1848 evaluated the preceding s-expression, which in this case was
1851 Since @code{+} does not have a value bound to it, just the function
1852 definition, the error message reported that the symbol's value as a
1857 In GNU Emacs version 20 and before, your error message will say:
1860 Symbol's value as variable is void:@: +
1864 The meaning is the same as in GNU Emacs 22.
1870 @cindex Passing information to functions
1872 To see how information is passed to functions, let's look again at
1873 our old standby, the addition of two plus two. In Lisp, this is written
1880 If you evaluate this expression, the number 4 will appear in your echo
1881 area. What the Lisp interpreter does is add the numbers that follow
1884 @cindex @samp{argument} defined
1885 The numbers added by @code{+} are called the @dfn{arguments} of the
1886 function @code{+}. These numbers are the information that is given to
1887 or @dfn{passed} to the function.
1889 The word `argument' comes from the way it is used in mathematics and
1890 does not refer to a disputation between two people; instead it refers to
1891 the information presented to the function, in this case, to the
1892 @code{+}. In Lisp, the arguments to a function are the atoms or lists
1893 that follow the function. The values returned by the evaluation of
1894 these atoms or lists are passed to the function. Different functions
1895 require different numbers of arguments; some functions require none at
1896 all.@footnote{It is curious to track the path by which the word `argument'
1897 came to have two different meanings, one in mathematics and the other in
1898 everyday English. According to the @cite{Oxford English Dictionary},
1899 the word derives from the Latin for @samp{to make clear, prove}; thus it
1900 came to mean, by one thread of derivation, `the evidence offered as
1901 proof', which is to say, `the information offered', which led to its
1902 meaning in Lisp. But in the other thread of derivation, it came to mean
1903 `to assert in a manner against which others may make counter
1904 assertions', which led to the meaning of the word as a disputation.
1905 (Note here that the English word has two different definitions attached
1906 to it at the same time. By contrast, in Emacs Lisp, a symbol cannot
1907 have two different function definitions at the same time.)}
1910 * Data types:: Types of data passed to a function.
1911 * Args as Variable or List:: An argument can be the value
1912 of a variable or list.
1913 * Variable Number of Arguments:: Some functions may take a
1914 variable number of arguments.
1915 * Wrong Type of Argument:: Passing an argument of the wrong type
1917 * message:: A useful function for sending messages.
1921 @subsection Arguments' Data Types
1923 @cindex Types of data
1924 @cindex Arguments' data types
1926 The type of data that should be passed to a function depends on what
1927 kind of information it uses. The arguments to a function such as
1928 @code{+} must have values that are numbers, since @code{+} adds numbers.
1929 Other functions use different kinds of data for their arguments.
1933 For example, the @code{concat} function links together or unites two or
1934 more strings of text to produce a string. The arguments are strings.
1935 Concatenating the two character strings @code{abc}, @code{def} produces
1936 the single string @code{abcdef}. This can be seen by evaluating the
1940 (concat "abc" "def")
1944 The value produced by evaluating this expression is @code{"abcdef"}.
1946 A function such as @code{substring} uses both a string and numbers as
1947 arguments. The function returns a part of the string, a substring of
1948 the first argument. This function takes three arguments. Its first
1949 argument is the string of characters, the second and third arguments are
1950 numbers that indicate the beginning and end of the substring. The
1951 numbers are a count of the number of characters (including spaces and
1952 punctuation) from the beginning of the string.
1955 For example, if you evaluate the following:
1958 (substring "The quick brown fox jumped." 16 19)
1962 you will see @code{"fox"} appear in the echo area. The arguments are the
1963 string and the two numbers.
1965 Note that the string passed to @code{substring} is a single atom even
1966 though it is made up of several words separated by spaces. Lisp counts
1967 everything between the two quotation marks as part of the string,
1968 including the spaces. You can think of the @code{substring} function as
1969 a kind of `atom smasher' since it takes an otherwise indivisible atom
1970 and extracts a part. However, @code{substring} is only able to extract
1971 a substring from an argument that is a string, not from another type of
1972 atom such as a number or symbol.
1974 @node Args as Variable or List
1975 @subsection An Argument as the Value of a Variable or List
1977 An argument can be a symbol that returns a value when it is evaluated.
1978 For example, when the symbol @code{fill-column} by itself is evaluated,
1979 it returns a number. This number can be used in an addition.
1982 Position the cursor after the following expression and type @kbd{C-x
1990 The value will be a number two more than what you get by evaluating
1991 @code{fill-column} alone. For me, this is 74, because my value of
1992 @code{fill-column} is 72.
1994 As we have just seen, an argument can be a symbol that returns a value
1995 when evaluated. In addition, an argument can be a list that returns a
1996 value when it is evaluated. For example, in the following expression,
1997 the arguments to the function @code{concat} are the strings
1998 @w{@code{"The "}} and @w{@code{" red foxes."}} and the list
1999 @code{(number-to-string (+ 2 fill-column))}.
2001 @c For GNU Emacs 22, need number-to-string
2003 (concat "The " (number-to-string (+ 2 fill-column)) " red foxes.")
2007 If you evaluate this expression---and if, as with my Emacs,
2008 @code{fill-column} evaluates to 72---@code{"The 74 red foxes."} will
2009 appear in the echo area. (Note that you must put spaces after the
2010 word @samp{The} and before the word @samp{red} so they will appear in
2011 the final string. The function @code{number-to-string} converts the
2012 integer that the addition function returns to a string.
2013 @code{number-to-string} is also known as @code{int-to-string}.)
2015 @node Variable Number of Arguments
2016 @subsection Variable Number of Arguments
2017 @cindex Variable number of arguments
2018 @cindex Arguments, variable number of
2020 Some functions, such as @code{concat}, @code{+} or @code{*}, take any
2021 number of arguments. (The @code{*} is the symbol for multiplication.)
2022 This can be seen by evaluating each of the following expressions in
2023 the usual way. What you will see in the echo area is printed in this
2024 text after @samp{@result{}}, which you may read as `evaluates to'.
2027 In the first set, the functions have no arguments:
2038 In this set, the functions have one argument each:
2049 In this set, the functions have three arguments each:
2053 (+ 3 4 5) @result{} 12
2055 (* 3 4 5) @result{} 60
2059 @node Wrong Type of Argument
2060 @subsection Using the Wrong Type Object as an Argument
2061 @cindex Wrong type of argument
2062 @cindex Argument, wrong type of
2064 When a function is passed an argument of the wrong type, the Lisp
2065 interpreter produces an error message. For example, the @code{+}
2066 function expects the values of its arguments to be numbers. As an
2067 experiment we can pass it the quoted symbol @code{hello} instead of a
2068 number. Position the cursor after the following expression and type
2076 When you do this you will generate an error message. What has happened
2077 is that @code{+} has tried to add the 2 to the value returned by
2078 @code{'hello}, but the value returned by @code{'hello} is the symbol
2079 @code{hello}, not a number. Only numbers can be added. So @code{+}
2080 could not carry out its addition.
2083 You will create and enter a @file{*Backtrace*} buffer that says:
2088 ---------- Buffer: *Backtrace* ----------
2089 Debugger entered--Lisp error:
2090 (wrong-type-argument number-or-marker-p hello)
2092 eval((+ 2 (quote hello)))
2093 eval-last-sexp-1(nil)
2095 call-interactively(eval-last-sexp)
2096 ---------- Buffer: *Backtrace* ----------
2101 As usual, the error message tries to be helpful and makes sense after you
2102 learn how to read it.@footnote{@code{(quote hello)} is an expansion of
2103 the abbreviation @code{'hello}.}
2105 The first part of the error message is straightforward; it says
2106 @samp{wrong type argument}. Next comes the mysterious jargon word
2107 @w{@samp{number-or-marker-p}}. This word is trying to tell you what
2108 kind of argument the @code{+} expected.
2110 The symbol @code{number-or-marker-p} says that the Lisp interpreter is
2111 trying to determine whether the information presented it (the value of
2112 the argument) is a number or a marker (a special object representing a
2113 buffer position). What it does is test to see whether the @code{+} is
2114 being given numbers to add. It also tests to see whether the
2115 argument is something called a marker, which is a specific feature of
2116 Emacs Lisp. (In Emacs, locations in a buffer are recorded as markers.
2117 When the mark is set with the @kbd{C-@@} or @kbd{C-@key{SPC}} command,
2118 its position is kept as a marker. The mark can be considered a
2119 number---the number of characters the location is from the beginning
2120 of the buffer.) In Emacs Lisp, @code{+} can be used to add the
2121 numeric value of marker positions as numbers.
2123 The @samp{p} of @code{number-or-marker-p} is the embodiment of a
2124 practice started in the early days of Lisp programming. The @samp{p}
2125 stands for `predicate'. In the jargon used by the early Lisp
2126 researchers, a predicate refers to a function to determine whether some
2127 property is true or false. So the @samp{p} tells us that
2128 @code{number-or-marker-p} is the name of a function that determines
2129 whether it is true or false that the argument supplied is a number or
2130 a marker. Other Lisp symbols that end in @samp{p} include @code{zerop},
2131 a function that tests whether its argument has the value of zero, and
2132 @code{listp}, a function that tests whether its argument is a list.
2134 Finally, the last part of the error message is the symbol @code{hello}.
2135 This is the value of the argument that was passed to @code{+}. If the
2136 addition had been passed the correct type of object, the value passed
2137 would have been a number, such as 37, rather than a symbol like
2138 @code{hello}. But then you would not have got the error message.
2142 In GNU Emacs version 20 and before, the echo area displays an error
2146 Wrong type argument:@: number-or-marker-p, hello
2149 This says, in different words, the same as the top line of the
2150 @file{*Backtrace*} buffer.
2154 @subsection The @code{message} Function
2157 Like @code{+}, the @code{message} function takes a variable number of
2158 arguments. It is used to send messages to the user and is so useful
2159 that we will describe it here.
2162 A message is printed in the echo area. For example, you can print a
2163 message in your echo area by evaluating the following list:
2166 (message "This message appears in the echo area!")
2169 The whole string between double quotation marks is a single argument
2170 and is printed @i{in toto}. (Note that in this example, the message
2171 itself will appear in the echo area within double quotes; that is
2172 because you see the value returned by the @code{message} function. In
2173 most uses of @code{message} in programs that you write, the text will
2174 be printed in the echo area as a side-effect, without the quotes.
2175 @xref{multiply-by-seven in detail, , @code{multiply-by-seven} in
2176 detail}, for an example of this.)
2178 However, if there is a @samp{%s} in the quoted string of characters, the
2179 @code{message} function does not print the @samp{%s} as such, but looks
2180 to the argument that follows the string. It evaluates the second
2181 argument and prints the value at the location in the string where the
2185 You can see this by positioning the cursor after the following
2186 expression and typing @kbd{C-x C-e}:
2189 (message "The name of this buffer is: %s." (buffer-name))
2193 In Info, @code{"The name of this buffer is: *info*."} will appear in the
2194 echo area. The function @code{buffer-name} returns the name of the
2195 buffer as a string, which the @code{message} function inserts in place
2198 To print a value as an integer, use @samp{%d} in the same way as
2199 @samp{%s}. For example, to print a message in the echo area that
2200 states the value of the @code{fill-column}, evaluate the following:
2203 (message "The value of fill-column is %d." fill-column)
2207 On my system, when I evaluate this list, @code{"The value of
2208 fill-column is 72."} appears in my echo area@footnote{Actually, you
2209 can use @code{%s} to print a number. It is non-specific. @code{%d}
2210 prints only the part of a number left of a decimal point, and not
2211 anything that is not a number.}.
2213 If there is more than one @samp{%s} in the quoted string, the value of
2214 the first argument following the quoted string is printed at the
2215 location of the first @samp{%s} and the value of the second argument is
2216 printed at the location of the second @samp{%s}, and so on.
2219 For example, if you evaluate the following,
2223 (message "There are %d %s in the office!"
2224 (- fill-column 14) "pink elephants")
2229 a rather whimsical message will appear in your echo area. On my system
2230 it says, @code{"There are 58 pink elephants in the office!"}.
2232 The expression @code{(- fill-column 14)} is evaluated and the resulting
2233 number is inserted in place of the @samp{%d}; and the string in double
2234 quotes, @code{"pink elephants"}, is treated as a single argument and
2235 inserted in place of the @samp{%s}. (That is to say, a string between
2236 double quotes evaluates to itself, like a number.)
2238 Finally, here is a somewhat complex example that not only illustrates
2239 the computation of a number, but also shows how you can use an
2240 expression within an expression to generate the text that is substituted
2245 (message "He saw %d %s"
2249 "The quick brown foxes jumped." 16 21)
2254 In this example, @code{message} has three arguments: the string,
2255 @code{"He saw %d %s"}, the expression, @code{(- fill-column 32)}, and
2256 the expression beginning with the function @code{concat}. The value
2257 resulting from the evaluation of @code{(- fill-column 32)} is inserted
2258 in place of the @samp{%d}; and the value returned by the expression
2259 beginning with @code{concat} is inserted in place of the @samp{%s}.
2261 When your fill column is 70 and you evaluate the expression, the
2262 message @code{"He saw 38 red foxes leaping."} appears in your echo
2266 @section Setting the Value of a Variable
2267 @cindex Variable, setting value
2268 @cindex Setting value of variable
2270 @cindex @samp{bind} defined
2271 There are several ways by which a variable can be given a value. One of
2272 the ways is to use either the function @code{set} or the function
2273 @code{setq}. Another way is to use @code{let} (@pxref{let}). (The
2274 jargon for this process is to @dfn{bind} a variable to a value.)
2276 The following sections not only describe how @code{set} and @code{setq}
2277 work but also illustrate how arguments are passed.
2280 * Using set:: Setting values.
2281 * Using setq:: Setting a quoted value.
2282 * Counting:: Using @code{setq} to count.
2286 @subsection Using @code{set}
2289 To set the value of the symbol @code{flowers} to the list @code{'(rose
2290 violet daisy buttercup)}, evaluate the following expression by
2291 positioning the cursor after the expression and typing @kbd{C-x C-e}.
2294 (set 'flowers '(rose violet daisy buttercup))
2298 The list @code{(rose violet daisy buttercup)} will appear in the echo
2299 area. This is what is @emph{returned} by the @code{set} function. As a
2300 side effect, the symbol @code{flowers} is bound to the list; that is,
2301 the symbol @code{flowers}, which can be viewed as a variable, is given
2302 the list as its value. (This process, by the way, illustrates how a
2303 side effect to the Lisp interpreter, setting the value, can be the
2304 primary effect that we humans are interested in. This is because every
2305 Lisp function must return a value if it does not get an error, but it
2306 will only have a side effect if it is designed to have one.)
2308 After evaluating the @code{set} expression, you can evaluate the symbol
2309 @code{flowers} and it will return the value you just set. Here is the
2310 symbol. Place your cursor after it and type @kbd{C-x C-e}.
2317 When you evaluate @code{flowers}, the list
2318 @code{(rose violet daisy buttercup)} appears in the echo area.
2320 Incidentally, if you evaluate @code{'flowers}, the variable with a quote
2321 in front of it, what you will see in the echo area is the symbol itself,
2322 @code{flowers}. Here is the quoted symbol, so you can try this:
2328 Note also, that when you use @code{set}, you need to quote both
2329 arguments to @code{set}, unless you want them evaluated. Since we do
2330 not want either argument evaluated, neither the variable
2331 @code{flowers} nor the list @code{(rose violet daisy buttercup)}, both
2332 are quoted. (When you use @code{set} without quoting its first
2333 argument, the first argument is evaluated before anything else is
2334 done. If you did this and @code{flowers} did not have a value
2335 already, you would get an error message that the @samp{Symbol's value
2336 as variable is void}; on the other hand, if @code{flowers} did return
2337 a value after it was evaluated, the @code{set} would attempt to set
2338 the value that was returned. There are situations where this is the
2339 right thing for the function to do; but such situations are rare.)
2342 @subsection Using @code{setq}
2345 As a practical matter, you almost always quote the first argument to
2346 @code{set}. The combination of @code{set} and a quoted first argument
2347 is so common that it has its own name: the special form @code{setq}.
2348 This special form is just like @code{set} except that the first argument
2349 is quoted automatically, so you don't need to type the quote mark
2350 yourself. Also, as an added convenience, @code{setq} permits you to set
2351 several different variables to different values, all in one expression.
2353 To set the value of the variable @code{carnivores} to the list
2354 @code{'(lion tiger leopard)} using @code{setq}, the following expression
2358 (setq carnivores '(lion tiger leopard))
2362 This is exactly the same as using @code{set} except the first argument
2363 is automatically quoted by @code{setq}. (The @samp{q} in @code{setq}
2364 means @code{quote}.)
2367 With @code{set}, the expression would look like this:
2370 (set 'carnivores '(lion tiger leopard))
2373 Also, @code{setq} can be used to assign different values to
2374 different variables. The first argument is bound to the value
2375 of the second argument, the third argument is bound to the value of the
2376 fourth argument, and so on. For example, you could use the following to
2377 assign a list of trees to the symbol @code{trees} and a list of herbivores
2378 to the symbol @code{herbivores}:
2382 (setq trees '(pine fir oak maple)
2383 herbivores '(gazelle antelope zebra))
2388 (The expression could just as well have been on one line, but it might
2389 not have fit on a page; and humans find it easier to read nicely
2392 Although I have been using the term `assign', there is another way of
2393 thinking about the workings of @code{set} and @code{setq}; and that is to
2394 say that @code{set} and @code{setq} make the symbol @emph{point} to the
2395 list. This latter way of thinking is very common and in forthcoming
2396 chapters we shall come upon at least one symbol that has `pointer' as
2397 part of its name. The name is chosen because the symbol has a value,
2398 specifically a list, attached to it; or, expressed another way,
2399 the symbol is set to ``point'' to the list.
2402 @subsection Counting
2405 Here is an example that shows how to use @code{setq} in a counter. You
2406 might use this to count how many times a part of your program repeats
2407 itself. First set a variable to zero; then add one to the number each
2408 time the program repeats itself. To do this, you need a variable that
2409 serves as a counter, and two expressions: an initial @code{setq}
2410 expression that sets the counter variable to zero; and a second
2411 @code{setq} expression that increments the counter each time it is
2416 (setq counter 0) ; @r{Let's call this the initializer.}
2418 (setq counter (+ counter 1)) ; @r{This is the incrementer.}
2420 counter ; @r{This is the counter.}
2425 (The text following the @samp{;} are comments. @xref{Change a
2426 defun, , Change a Function Definition}.)
2428 If you evaluate the first of these expressions, the initializer,
2429 @code{(setq counter 0)}, and then evaluate the third expression,
2430 @code{counter}, the number @code{0} will appear in the echo area. If
2431 you then evaluate the second expression, the incrementer, @code{(setq
2432 counter (+ counter 1))}, the counter will get the value 1. So if you
2433 again evaluate @code{counter}, the number @code{1} will appear in the
2434 echo area. Each time you evaluate the second expression, the value of
2435 the counter will be incremented.
2437 When you evaluate the incrementer, @code{(setq counter (+ counter 1))},
2438 the Lisp interpreter first evaluates the innermost list; this is the
2439 addition. In order to evaluate this list, it must evaluate the variable
2440 @code{counter} and the number @code{1}. When it evaluates the variable
2441 @code{counter}, it receives its current value. It passes this value and
2442 the number @code{1} to the @code{+} which adds them together. The sum
2443 is then returned as the value of the inner list and passed to the
2444 @code{setq} which sets the variable @code{counter} to this new value.
2445 Thus, the value of the variable, @code{counter}, is changed.
2450 Learning Lisp is like climbing a hill in which the first part is the
2451 steepest. You have now climbed the most difficult part; what remains
2452 becomes easier as you progress onwards.
2460 Lisp programs are made up of expressions, which are lists or single atoms.
2463 Lists are made up of zero or more atoms or inner lists, separated by whitespace and
2464 surrounded by parentheses. A list can be empty.
2467 Atoms are multi-character symbols, like @code{forward-paragraph}, single
2468 character symbols like @code{+}, strings of characters between double
2469 quotation marks, or numbers.
2472 A number evaluates to itself.
2475 A string between double quotes also evaluates to itself.
2478 When you evaluate a symbol by itself, its value is returned.
2481 When you evaluate a list, the Lisp interpreter looks at the first symbol
2482 in the list and then at the function definition bound to that symbol.
2483 Then the instructions in the function definition are carried out.
2486 A single quotation mark,
2493 , tells the Lisp interpreter that it should
2494 return the following expression as written, and not evaluate it as it
2495 would if the quote were not there.
2498 Arguments are the information passed to a function. The arguments to a
2499 function are computed by evaluating the rest of the elements of the list
2500 of which the function is the first element.
2503 A function always returns a value when it is evaluated (unless it gets
2504 an error); in addition, it may also carry out some action called a
2505 ``side effect''. In many cases, a function's primary purpose is to
2506 create a side effect.
2509 @node Error Message Exercises
2512 A few simple exercises:
2516 Generate an error message by evaluating an appropriate symbol that is
2517 not within parentheses.
2520 Generate an error message by evaluating an appropriate symbol that is
2521 between parentheses.
2524 Create a counter that increments by two rather than one.
2527 Write an expression that prints a message in the echo area when
2531 @node Practicing Evaluation
2532 @chapter Practicing Evaluation
2533 @cindex Practicing evaluation
2534 @cindex Evaluation practice
2536 Before learning how to write a function definition in Emacs Lisp, it is
2537 useful to spend a little time evaluating various expressions that have
2538 already been written. These expressions will be lists with the
2539 functions as their first (and often only) element. Since some of the
2540 functions associated with buffers are both simple and interesting, we
2541 will start with those. In this section, we will evaluate a few of
2542 these. In another section, we will study the code of several other
2543 buffer-related functions, to see how they were written.
2546 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
2548 * Buffer Names:: Buffers and files are different.
2549 * Getting Buffers:: Getting a buffer itself, not merely its name.
2550 * Switching Buffers:: How to change to another buffer.
2551 * Buffer Size & Locations:: Where point is located and the size of
2553 * Evaluation Exercise::
2557 @node How to Evaluate
2558 @unnumberedsec How to Evaluate
2561 @i{Whenever you give an editing command} to Emacs Lisp, such as the
2562 command to move the cursor or to scroll the screen, @i{you are evaluating
2563 an expression,} the first element of which is a function. @i{This is
2566 @cindex @samp{interactive function} defined
2567 @cindex @samp{command} defined
2568 When you type keys, you cause the Lisp interpreter to evaluate an
2569 expression and that is how you get your results. Even typing plain text
2570 involves evaluating an Emacs Lisp function, in this case, one that uses
2571 @code{self-insert-command}, which simply inserts the character you
2572 typed. The functions you evaluate by typing keystrokes are called
2573 @dfn{interactive} functions, or @dfn{commands}; how you make a function
2574 interactive will be illustrated in the chapter on how to write function
2575 definitions. @xref{Interactive, , Making a Function Interactive}.
2577 In addition to typing keyboard commands, we have seen a second way to
2578 evaluate an expression: by positioning the cursor after a list and
2579 typing @kbd{C-x C-e}. This is what we will do in the rest of this
2580 section. There are other ways to evaluate an expression as well; these
2581 will be described as we come to them.
2583 Besides being used for practicing evaluation, the functions shown in the
2584 next few sections are important in their own right. A study of these
2585 functions makes clear the distinction between buffers and files, how to
2586 switch to a buffer, and how to determine a location within it.
2589 @section Buffer Names
2591 @findex buffer-file-name
2593 The two functions, @code{buffer-name} and @code{buffer-file-name}, show
2594 the difference between a file and a buffer. When you evaluate the
2595 following expression, @code{(buffer-name)}, the name of the buffer
2596 appears in the echo area. When you evaluate @code{(buffer-file-name)},
2597 the name of the file to which the buffer refers appears in the echo
2598 area. Usually, the name returned by @code{(buffer-name)} is the same as
2599 the name of the file to which it refers, and the name returned by
2600 @code{(buffer-file-name)} is the full path-name of the file.
2602 A file and a buffer are two different entities. A file is information
2603 recorded permanently in the computer (unless you delete it). A buffer,
2604 on the other hand, is information inside of Emacs that will vanish at
2605 the end of the editing session (or when you kill the buffer). Usually,
2606 a buffer contains information that you have copied from a file; we say
2607 the buffer is @dfn{visiting} that file. This copy is what you work on
2608 and modify. Changes to the buffer do not change the file, until you
2609 save the buffer. When you save the buffer, the buffer is copied to the file
2610 and is thus saved permanently.
2613 If you are reading this in Info inside of GNU Emacs, you can evaluate
2614 each of the following expressions by positioning the cursor after it and
2615 typing @kbd{C-x C-e}.
2626 When I do this in Info, the value returned by evaluating
2627 @code{(buffer-name)} is @file{"*info*"}, and the value returned by
2628 evaluating @code{(buffer-file-name)} is @file{nil}.
2630 On the other hand, while I am writing this document, the value
2631 returned by evaluating @code{(buffer-name)} is
2632 @file{"introduction.texinfo"}, and the value returned by evaluating
2633 @code{(buffer-file-name)} is
2634 @file{"/gnu/work/intro/introduction.texinfo"}.
2636 @cindex @code{nil}, history of word
2637 The former is the name of the buffer and the latter is the name of the
2638 file. In Info, the buffer name is @file{"*info*"}. Info does not
2639 point to any file, so the result of evaluating
2640 @code{(buffer-file-name)} is @file{nil}. The symbol @code{nil} is
2641 from the Latin word for `nothing'; in this case, it means that the
2642 buffer is not associated with any file. (In Lisp, @code{nil} is also
2643 used to mean `false' and is a synonym for the empty list, @code{()}.)
2645 When I am writing, the name of my buffer is
2646 @file{"introduction.texinfo"}. The name of the file to which it
2647 points is @file{"/gnu/work/intro/introduction.texinfo"}.
2649 (In the expressions, the parentheses tell the Lisp interpreter to
2650 treat @w{@code{buffer-name}} and @w{@code{buffer-file-name}} as
2651 functions; without the parentheses, the interpreter would attempt to
2652 evaluate the symbols as variables. @xref{Variables}.)
2654 In spite of the distinction between files and buffers, you will often
2655 find that people refer to a file when they mean a buffer and vice-verse.
2656 Indeed, most people say, ``I am editing a file,'' rather than saying,
2657 ``I am editing a buffer which I will soon save to a file.'' It is
2658 almost always clear from context what people mean. When dealing with
2659 computer programs, however, it is important to keep the distinction in mind,
2660 since the computer is not as smart as a person.
2662 @cindex Buffer, history of word
2663 The word `buffer', by the way, comes from the meaning of the word as a
2664 cushion that deadens the force of a collision. In early computers, a
2665 buffer cushioned the interaction between files and the computer's
2666 central processing unit. The drums or tapes that held a file and the
2667 central processing unit were pieces of equipment that were very
2668 different from each other, working at their own speeds, in spurts. The
2669 buffer made it possible for them to work together effectively.
2670 Eventually, the buffer grew from being an intermediary, a temporary
2671 holding place, to being the place where work is done. This
2672 transformation is rather like that of a small seaport that grew into a
2673 great city: once it was merely the place where cargo was warehoused
2674 temporarily before being loaded onto ships; then it became a business
2675 and cultural center in its own right.
2677 Not all buffers are associated with files. For example, a
2678 @file{*scratch*} buffer does not visit any file. Similarly, a
2679 @file{*Help*} buffer is not associated with any file.
2681 In the old days, when you lacked a @file{~/.emacs} file and started an
2682 Emacs session by typing the command @code{emacs} alone, without naming
2683 any files, Emacs started with the @file{*scratch*} buffer visible.
2684 Nowadays, you will see a splash screen. You can follow one of the
2685 commands suggested on the splash screen, visit a file, or press the
2686 spacebar to reach the @file{*scratch*} buffer.
2688 If you switch to the @file{*scratch*} buffer, type
2689 @code{(buffer-name)}, position the cursor after it, and then type
2690 @kbd{C-x C-e} to evaluate the expression. The name @code{"*scratch*"}
2691 will be returned and will appear in the echo area. @code{"*scratch*"}
2692 is the name of the buffer. When you type @code{(buffer-file-name)} in
2693 the @file{*scratch*} buffer and evaluate that, @code{nil} will appear
2694 in the echo area, just as it does when you evaluate
2695 @code{(buffer-file-name)} in Info.
2697 Incidentally, if you are in the @file{*scratch*} buffer and want the
2698 value returned by an expression to appear in the @file{*scratch*}
2699 buffer itself rather than in the echo area, type @kbd{C-u C-x C-e}
2700 instead of @kbd{C-x C-e}. This causes the value returned to appear
2701 after the expression. The buffer will look like this:
2704 (buffer-name)"*scratch*"
2708 You cannot do this in Info since Info is read-only and it will not allow
2709 you to change the contents of the buffer. But you can do this in any
2710 buffer you can edit; and when you write code or documentation (such as
2711 this book), this feature is very useful.
2713 @node Getting Buffers
2714 @section Getting Buffers
2715 @findex current-buffer
2716 @findex other-buffer
2717 @cindex Getting a buffer
2719 The @code{buffer-name} function returns the @emph{name} of the buffer;
2720 to get the buffer @emph{itself}, a different function is needed: the
2721 @code{current-buffer} function. If you use this function in code, what
2722 you get is the buffer itself.
2724 A name and the object or entity to which the name refers are different
2725 from each other. You are not your name. You are a person to whom
2726 others refer by name. If you ask to speak to George and someone hands you
2727 a card with the letters @samp{G}, @samp{e}, @samp{o}, @samp{r},
2728 @samp{g}, and @samp{e} written on it, you might be amused, but you would
2729 not be satisfied. You do not want to speak to the name, but to the
2730 person to whom the name refers. A buffer is similar: the name of the
2731 scratch buffer is @file{*scratch*}, but the name is not the buffer. To
2732 get a buffer itself, you need to use a function such as
2733 @code{current-buffer}.
2735 However, there is a slight complication: if you evaluate
2736 @code{current-buffer} in an expression on its own, as we will do here,
2737 what you see is a printed representation of the name of the buffer
2738 without the contents of the buffer. Emacs works this way for two
2739 reasons: the buffer may be thousands of lines long---too long to be
2740 conveniently displayed; and, another buffer may have the same contents
2741 but a different name, and it is important to distinguish between them.
2744 Here is an expression containing the function:
2751 If you evaluate this expression in Info in Emacs in the usual way,
2752 @file{#<buffer *info*>} will appear in the echo area. The special
2753 format indicates that the buffer itself is being returned, rather than
2756 Incidentally, while you can type a number or symbol into a program, you
2757 cannot do that with the printed representation of a buffer: the only way
2758 to get a buffer itself is with a function such as @code{current-buffer}.
2760 A related function is @code{other-buffer}. This returns the most
2761 recently selected buffer other than the one you are in currently, not
2762 a printed representation of its name. If you have recently switched
2763 back and forth from the @file{*scratch*} buffer, @code{other-buffer}
2764 will return that buffer.
2767 You can see this by evaluating the expression:
2774 You should see @file{#<buffer *scratch*>} appear in the echo area, or
2775 the name of whatever other buffer you switched back from most
2776 recently@footnote{Actually, by default, if the buffer from which you
2777 just switched is visible to you in another window, @code{other-buffer}
2778 will choose the most recent buffer that you cannot see; this is a
2779 subtlety that I often forget.}.
2781 @node Switching Buffers
2782 @section Switching Buffers
2783 @findex switch-to-buffer
2785 @cindex Switching to a buffer
2787 The @code{other-buffer} function actually provides a buffer when it is
2788 used as an argument to a function that requires one. We can see this
2789 by using @code{other-buffer} and @code{switch-to-buffer} to switch to a
2792 But first, a brief introduction to the @code{switch-to-buffer}
2793 function. When you switched back and forth from Info to the
2794 @file{*scratch*} buffer to evaluate @code{(buffer-name)}, you most
2795 likely typed @kbd{C-x b} and then typed @file{*scratch*}@footnote{Or
2796 rather, to save typing, you probably only typed @kbd{RET} if the
2797 default buffer was @file{*scratch*}, or if it was different, then you
2798 typed just part of the name, such as @code{*sc}, pressed your
2799 @kbd{TAB} key to cause it to expand to the full name, and then typed
2800 @kbd{RET}.} when prompted in the minibuffer for the name of
2801 the buffer to which you wanted to switch. The keystrokes, @kbd{C-x
2802 b}, cause the Lisp interpreter to evaluate the interactive function
2803 @code{switch-to-buffer}. As we said before, this is how Emacs works:
2804 different keystrokes call or run different functions. For example,
2805 @kbd{C-f} calls @code{forward-char}, @kbd{M-e} calls
2806 @code{forward-sentence}, and so on.
2808 By writing @code{switch-to-buffer} in an expression, and giving it a
2809 buffer to switch to, we can switch buffers just the way @kbd{C-x b}
2813 (switch-to-buffer (other-buffer))
2817 The symbol @code{switch-to-buffer} is the first element of the list,
2818 so the Lisp interpreter will treat it as a function and carry out the
2819 instructions that are attached to it. But before doing that, the
2820 interpreter will note that @code{other-buffer} is inside parentheses
2821 and work on that symbol first. @code{other-buffer} is the first (and
2822 in this case, the only) element of this list, so the Lisp interpreter
2823 calls or runs the function. It returns another buffer. Next, the
2824 interpreter runs @code{switch-to-buffer}, passing to it, as an
2825 argument, the other buffer, which is what Emacs will switch to. If
2826 you are reading this in Info, try this now. Evaluate the expression.
2827 (To get back, type @kbd{C-x b @key{RET}}.)@footnote{Remember, this
2828 expression will move you to your most recent other buffer that you
2829 cannot see. If you really want to go to your most recently selected
2830 buffer, even if you can still see it, you need to evaluate the
2831 following more complex expression:
2834 (switch-to-buffer (other-buffer (current-buffer) t))
2838 In this case, the first argument to @code{other-buffer} tells it which
2839 buffer to skip---the current one---and the second argument tells
2840 @code{other-buffer} it is OK to switch to a visible buffer.
2841 In regular use, @code{switch-to-buffer} takes you to an invisible
2842 window since you would most likely use @kbd{C-x o} (@code{other-window})
2843 to go to another visible buffer.}
2845 In the programming examples in later sections of this document, you will
2846 see the function @code{set-buffer} more often than
2847 @code{switch-to-buffer}. This is because of a difference between
2848 computer programs and humans: humans have eyes and expect to see the
2849 buffer on which they are working on their computer terminals. This is
2850 so obvious, it almost goes without saying. However, programs do not
2851 have eyes. When a computer program works on a buffer, that buffer does
2852 not need to be visible on the screen.
2854 @code{switch-to-buffer} is designed for humans and does two different
2855 things: it switches the buffer to which Emacs's attention is directed; and
2856 it switches the buffer displayed in the window to the new buffer.
2857 @code{set-buffer}, on the other hand, does only one thing: it switches
2858 the attention of the computer program to a different buffer. The buffer
2859 on the screen remains unchanged (of course, normally nothing happens
2860 there until the command finishes running).
2862 @cindex @samp{call} defined
2863 Also, we have just introduced another jargon term, the word @dfn{call}.
2864 When you evaluate a list in which the first symbol is a function, you
2865 are calling that function. The use of the term comes from the notion of
2866 the function as an entity that can do something for you if you `call'
2867 it---just as a plumber is an entity who can fix a leak if you call him
2870 @node Buffer Size & Locations
2871 @section Buffer Size and the Location of Point
2872 @cindex Size of buffer
2874 @cindex Point location
2875 @cindex Location of point
2877 Finally, let's look at several rather simple functions,
2878 @code{buffer-size}, @code{point}, @code{point-min}, and
2879 @code{point-max}. These give information about the size of a buffer and
2880 the location of point within it.
2882 The function @code{buffer-size} tells you the size of the current
2883 buffer; that is, the function returns a count of the number of
2884 characters in the buffer.
2891 You can evaluate this in the usual way, by positioning the
2892 cursor after the expression and typing @kbd{C-x C-e}.
2894 @cindex @samp{point} defined
2895 In Emacs, the current position of the cursor is called @dfn{point}.
2896 The expression @code{(point)} returns a number that tells you where the
2897 cursor is located as a count of the number of characters from the
2898 beginning of the buffer up to point.
2901 You can see the character count for point in this buffer by evaluating
2902 the following expression in the usual way:
2909 As I write this, the value of @code{point} is 65724. The @code{point}
2910 function is frequently used in some of the examples later in this
2914 The value of point depends, of course, on its location within the
2915 buffer. If you evaluate point in this spot, the number will be larger:
2922 For me, the value of point in this location is 66043, which means that
2923 there are 319 characters (including spaces) between the two
2924 expressions. (Doubtless, you will see different numbers, since I will
2925 have edited this since I first evaluated point.)
2927 @cindex @samp{narrowing} defined
2928 The function @code{point-min} is somewhat similar to @code{point}, but
2929 it returns the value of the minimum permissible value of point in the
2930 current buffer. This is the number 1 unless @dfn{narrowing} is in
2931 effect. (Narrowing is a mechanism whereby you can restrict yourself,
2932 or a program, to operations on just a part of a buffer.
2933 @xref{Narrowing & Widening, , Narrowing and Widening}.) Likewise, the
2934 function @code{point-max} returns the value of the maximum permissible
2935 value of point in the current buffer.
2937 @node Evaluation Exercise
2940 Find a file with which you are working and move towards its middle.
2941 Find its buffer name, file name, length, and your position in the file.
2943 @node Writing Defuns
2944 @chapter How To Write Function Definitions
2945 @cindex Definition writing
2946 @cindex Function definition writing
2947 @cindex Writing a function definition
2949 When the Lisp interpreter evaluates a list, it looks to see whether the
2950 first symbol on the list has a function definition attached to it; or,
2951 put another way, whether the symbol points to a function definition. If
2952 it does, the computer carries out the instructions in the definition. A
2953 symbol that has a function definition is called, simply, a function
2954 (although, properly speaking, the definition is the function and the
2955 symbol refers to it.)
2958 * Primitive Functions::
2959 * defun:: The @code{defun} macro.
2960 * Install:: Install a function definition.
2961 * Interactive:: Making a function interactive.
2962 * Interactive Options:: Different options for @code{interactive}.
2963 * Permanent Installation:: Installing code permanently.
2964 * let:: Creating and initializing local variables.
2966 * else:: If--then--else expressions.
2967 * Truth & Falsehood:: What Lisp considers false and true.
2968 * save-excursion:: Keeping track of point, mark, and buffer.
2974 @node Primitive Functions
2975 @unnumberedsec An Aside about Primitive Functions
2977 @cindex Primitive functions
2978 @cindex Functions, primitive
2980 @cindex C language primitives
2981 @cindex Primitives written in C
2982 All functions are defined in terms of other functions, except for a few
2983 @dfn{primitive} functions that are written in the C programming
2984 language. When you write functions' definitions, you will write them in
2985 Emacs Lisp and use other functions as your building blocks. Some of the
2986 functions you will use will themselves be written in Emacs Lisp (perhaps
2987 by you) and some will be primitives written in C@. The primitive
2988 functions are used exactly like those written in Emacs Lisp and behave
2989 like them. They are written in C so we can easily run GNU Emacs on any
2990 computer that has sufficient power and can run C.
2992 Let me re-emphasize this: when you write code in Emacs Lisp, you do not
2993 distinguish between the use of functions written in C and the use of
2994 functions written in Emacs Lisp. The difference is irrelevant. I
2995 mention the distinction only because it is interesting to know. Indeed,
2996 unless you investigate, you won't know whether an already-written
2997 function is written in Emacs Lisp or C.
3000 @section The @code{defun} Macro
3003 @cindex @samp{function definition} defined
3004 In Lisp, a symbol such as @code{mark-whole-buffer} has code attached to
3005 it that tells the computer what to do when the function is called.
3006 This code is called the @dfn{function definition} and is created by
3007 evaluating a Lisp expression that starts with the symbol @code{defun}
3008 (which is an abbreviation for @emph{define function}).
3010 In subsequent sections, we will look at function definitions from the
3011 Emacs source code, such as @code{mark-whole-buffer}. In this section,
3012 we will describe a simple function definition so you can see how it
3013 looks. This function definition uses arithmetic because it makes for a
3014 simple example. Some people dislike examples using arithmetic; however,
3015 if you are such a person, do not despair. Hardly any of the code we
3016 will study in the remainder of this introduction involves arithmetic or
3017 mathematics. The examples mostly involve text in one way or another.
3019 A function definition has up to five parts following the word
3024 The name of the symbol to which the function definition should be
3028 A list of the arguments that will be passed to the function. If no
3029 arguments will be passed to the function, this is an empty list,
3033 Documentation describing the function. (Technically optional, but
3034 strongly recommended.)
3037 Optionally, an expression to make the function interactive so you can
3038 use it by typing @kbd{M-x} and then the name of the function; or by
3039 typing an appropriate key or keychord.
3041 @cindex @samp{body} defined
3043 The code that instructs the computer what to do: the @dfn{body} of the
3044 function definition.
3047 It is helpful to think of the five parts of a function definition as
3048 being organized in a template, with slots for each part:
3052 (defun @var{function-name} (@var{arguments}@dots{})
3053 "@var{optional-documentation}@dots{}"
3054 (interactive @var{argument-passing-info}) ; @r{optional}
3059 As an example, here is the code for a function that multiplies its
3060 argument by 7. (This example is not interactive. @xref{Interactive,
3061 , Making a Function Interactive}, for that information.)
3065 (defun multiply-by-seven (number)
3066 "Multiply NUMBER by seven."
3071 This definition begins with a parenthesis and the symbol @code{defun},
3072 followed by the name of the function.
3074 @cindex @samp{argument list} defined
3075 The name of the function is followed by a list that contains the
3076 arguments that will be passed to the function. This list is called
3077 the @dfn{argument list}. In this example, the list has only one
3078 element, the symbol, @code{number}. When the function is used, the
3079 symbol will be bound to the value that is used as the argument to the
3082 Instead of choosing the word @code{number} for the name of the argument,
3083 I could have picked any other name. For example, I could have chosen
3084 the word @code{multiplicand}. I picked the word `number' because it
3085 tells what kind of value is intended for this slot; but I could just as
3086 well have chosen the word `multiplicand' to indicate the role that the
3087 value placed in this slot will play in the workings of the function. I
3088 could have called it @code{foogle}, but that would have been a bad
3089 choice because it would not tell humans what it means. The choice of
3090 name is up to the programmer and should be chosen to make the meaning of
3093 Indeed, you can choose any name you wish for a symbol in an argument
3094 list, even the name of a symbol used in some other function: the name
3095 you use in an argument list is private to that particular definition.
3096 In that definition, the name refers to a different entity than any use
3097 of the same name outside the function definition. Suppose you have a
3098 nick-name `Shorty' in your family; when your family members refer to
3099 `Shorty', they mean you. But outside your family, in a movie, for
3100 example, the name `Shorty' refers to someone else. Because a name in an
3101 argument list is private to the function definition, you can change the
3102 value of such a symbol inside the body of a function without changing
3103 its value outside the function. The effect is similar to that produced
3104 by a @code{let} expression. (@xref{let, , @code{let}}.)
3107 Note also that we discuss the word `number' in two different ways: as a
3108 symbol that appears in the code, and as the name of something that will
3109 be replaced by a something else during the evaluation of the function.
3110 In the first case, @code{number} is a symbol, not a number; it happens
3111 that within the function, it is a variable who value is the number in
3112 question, but our primary interest in it is as a symbol. On the other
3113 hand, when we are talking about the function, our interest is that we
3114 will substitute a number for the word @var{number}. To keep this
3115 distinction clear, we use different typography for the two
3116 circumstances. When we talk about this function, or about how it works,
3117 we refer to this number by writing @var{number}. In the function
3118 itself, we refer to it by writing @code{number}.
3121 The argument list is followed by the documentation string that
3122 describes the function. This is what you see when you type
3123 @w{@kbd{C-h f}} and the name of a function. Incidentally, when you
3124 write a documentation string like this, you should make the first line
3125 a complete sentence since some commands, such as @code{apropos}, print
3126 only the first line of a multi-line documentation string. Also, you
3127 should not indent the second line of a documentation string, if you
3128 have one, because that looks odd when you use @kbd{C-h f}
3129 (@code{describe-function}). The documentation string is optional, but
3130 it is so useful, it should be included in almost every function you
3133 @findex * @r{(multiplication)}
3134 The third line of the example consists of the body of the function
3135 definition. (Most functions' definitions, of course, are longer than
3136 this.) In this function, the body is the list, @code{(* 7 number)}, which
3137 says to multiply the value of @var{number} by 7. (In Emacs Lisp,
3138 @code{*} is the function for multiplication, just as @code{+} is the
3139 function for addition.)
3141 When you use the @code{multiply-by-seven} function, the argument
3142 @code{number} evaluates to the actual number you want used. Here is an
3143 example that shows how @code{multiply-by-seven} is used; but don't try
3144 to evaluate this yet!
3147 (multiply-by-seven 3)
3151 The symbol @code{number}, specified in the function definition in the
3152 next section, is given or ``bound to'' the value 3 in the actual use of
3153 the function. Note that although @code{number} was inside parentheses
3154 in the function definition, the argument passed to the
3155 @code{multiply-by-seven} function is not in parentheses. The
3156 parentheses are written in the function definition so the computer can
3157 figure out where the argument list ends and the rest of the function
3160 If you evaluate this example, you are likely to get an error message.
3161 (Go ahead, try it!) This is because we have written the function
3162 definition, but not yet told the computer about the definition---we have
3163 not yet installed (or `loaded') the function definition in Emacs.
3164 Installing a function is the process that tells the Lisp interpreter the
3165 definition of the function. Installation is described in the next
3169 @section Install a Function Definition
3170 @cindex Install a Function Definition
3171 @cindex Definition installation
3172 @cindex Function definition installation
3174 If you are reading this inside of Info in Emacs, you can try out the
3175 @code{multiply-by-seven} function by first evaluating the function
3176 definition and then evaluating @code{(multiply-by-seven 3)}. A copy of
3177 the function definition follows. Place the cursor after the last
3178 parenthesis of the function definition and type @kbd{C-x C-e}. When you
3179 do this, @code{multiply-by-seven} will appear in the echo area. (What
3180 this means is that when a function definition is evaluated, the value it
3181 returns is the name of the defined function.) At the same time, this
3182 action installs the function definition.
3186 (defun multiply-by-seven (number)
3187 "Multiply NUMBER by seven."
3193 By evaluating this @code{defun}, you have just installed
3194 @code{multiply-by-seven} in Emacs. The function is now just as much a
3195 part of Emacs as @code{forward-word} or any other editing function you
3196 use. (@code{multiply-by-seven} will stay installed until you quit
3197 Emacs. To reload code automatically whenever you start Emacs, see
3198 @ref{Permanent Installation, , Installing Code Permanently}.)
3201 * Effect of installation::
3202 * Change a defun:: How to change a function definition.
3206 @node Effect of installation
3207 @unnumberedsubsec The effect of installation
3210 You can see the effect of installing @code{multiply-by-seven} by
3211 evaluating the following sample. Place the cursor after the following
3212 expression and type @kbd{C-x C-e}. The number 21 will appear in the
3216 (multiply-by-seven 3)
3219 If you wish, you can read the documentation for the function by typing
3220 @kbd{C-h f} (@code{describe-function}) and then the name of the
3221 function, @code{multiply-by-seven}. When you do this, a
3222 @file{*Help*} window will appear on your screen that says:
3226 multiply-by-seven is a Lisp function.
3227 (multiply-by-seven NUMBER)
3229 Multiply NUMBER by seven.
3234 (To return to a single window on your screen, type @kbd{C-x 1}.)
3236 @node Change a defun
3237 @subsection Change a Function Definition
3238 @cindex Changing a function definition
3239 @cindex Function definition, how to change
3240 @cindex Definition, how to change
3242 If you want to change the code in @code{multiply-by-seven}, just rewrite
3243 it. To install the new version in place of the old one, evaluate the
3244 function definition again. This is how you modify code in Emacs. It is
3247 As an example, you can change the @code{multiply-by-seven} function to
3248 add the number to itself seven times instead of multiplying the number
3249 by seven. It produces the same answer, but by a different path. At
3250 the same time, we will add a comment to the code; a comment is text
3251 that the Lisp interpreter ignores, but that a human reader may find
3252 useful or enlightening. The comment is that this is the ``second
3257 (defun multiply-by-seven (number) ; @r{Second version.}
3258 "Multiply NUMBER by seven."
3259 (+ number number number number number number number))
3263 @cindex Comments in Lisp code
3264 The comment follows a semicolon, @samp{;}. In Lisp, everything on a
3265 line that follows a semicolon is a comment. The end of the line is the
3266 end of the comment. To stretch a comment over two or more lines, begin
3267 each line with a semicolon.
3269 @xref{Beginning init File, , Beginning a @file{.emacs}
3270 File}, and @ref{Comments, , Comments, elisp, The GNU Emacs Lisp
3271 Reference Manual}, for more about comments.
3273 You can install this version of the @code{multiply-by-seven} function by
3274 evaluating it in the same way you evaluated the first function: place
3275 the cursor after the last parenthesis and type @kbd{C-x C-e}.
3277 In summary, this is how you write code in Emacs Lisp: you write a
3278 function; install it; test it; and then make fixes or enhancements and
3282 @section Make a Function Interactive
3283 @cindex Interactive functions
3286 You make a function interactive by placing a list that begins with
3287 the special form @code{interactive} immediately after the
3288 documentation. A user can invoke an interactive function by typing
3289 @kbd{M-x} and then the name of the function; or by typing the keys to
3290 which it is bound, for example, by typing @kbd{C-n} for
3291 @code{next-line} or @kbd{C-x h} for @code{mark-whole-buffer}.
3293 Interestingly, when you call an interactive function interactively,
3294 the value returned is not automatically displayed in the echo area.
3295 This is because you often call an interactive function for its side
3296 effects, such as moving forward by a word or line, and not for the
3297 value returned. If the returned value were displayed in the echo area
3298 each time you typed a key, it would be very distracting.
3301 * Interactive multiply-by-seven:: An overview.
3302 * multiply-by-seven in detail:: The interactive version.
3306 @node Interactive multiply-by-seven
3307 @unnumberedsubsec An Interactive @code{multiply-by-seven}, An Overview
3310 Both the use of the special form @code{interactive} and one way to
3311 display a value in the echo area can be illustrated by creating an
3312 interactive version of @code{multiply-by-seven}.
3319 (defun multiply-by-seven (number) ; @r{Interactive version.}
3320 "Multiply NUMBER by seven."
3322 (message "The result is %d" (* 7 number)))
3327 You can install this code by placing your cursor after it and typing
3328 @kbd{C-x C-e}. The name of the function will appear in your echo area.
3329 Then, you can use this code by typing @kbd{C-u} and a number and then
3330 typing @kbd{M-x multiply-by-seven} and pressing @key{RET}. The phrase
3331 @samp{The result is @dots{}} followed by the product will appear in the
3334 Speaking more generally, you invoke a function like this in either of two
3339 By typing a prefix argument that contains the number to be passed, and
3340 then typing @kbd{M-x} and the name of the function, as with
3341 @kbd{C-u 3 M-x forward-sentence}; or,
3344 By typing whatever key or keychord the function is bound to, as with
3349 Both the examples just mentioned work identically to move point forward
3350 three sentences. (Since @code{multiply-by-seven} is not bound to a key,
3351 it could not be used as an example of key binding.)
3353 (@xref{Keybindings, , Some Keybindings}, to learn how to bind a command
3356 A prefix argument is passed to an interactive function by typing the
3357 @key{META} key followed by a number, for example, @kbd{M-3 M-e}, or by
3358 typing @kbd{C-u} and then a number, for example, @kbd{C-u 3 M-e} (if you
3359 type @kbd{C-u} without a number, it defaults to 4).
3361 @node multiply-by-seven in detail
3362 @subsection An Interactive @code{multiply-by-seven}
3364 Let's look at the use of the special form @code{interactive} and then at
3365 the function @code{message} in the interactive version of
3366 @code{multiply-by-seven}. You will recall that the function definition
3371 (defun multiply-by-seven (number) ; @r{Interactive version.}
3372 "Multiply NUMBER by seven."
3374 (message "The result is %d" (* 7 number)))
3378 In this function, the expression, @code{(interactive "p")}, is a list of
3379 two elements. The @code{"p"} tells Emacs to pass the prefix argument to
3380 the function and use its value for the argument of the function.
3383 The argument will be a number. This means that the symbol
3384 @code{number} will be bound to a number in the line:
3387 (message "The result is %d" (* 7 number))
3392 For example, if your prefix argument is 5, the Lisp interpreter will
3393 evaluate the line as if it were:
3396 (message "The result is %d" (* 7 5))
3400 (If you are reading this in GNU Emacs, you can evaluate this expression
3401 yourself.) First, the interpreter will evaluate the inner list, which
3402 is @code{(* 7 5)}. This returns a value of 35. Next, it
3403 will evaluate the outer list, passing the values of the second and
3404 subsequent elements of the list to the function @code{message}.
3406 As we have seen, @code{message} is an Emacs Lisp function especially
3407 designed for sending a one line message to a user. (@xref{message, ,
3408 The @code{message} function}.) In summary, the @code{message}
3409 function prints its first argument in the echo area as is, except for
3410 occurrences of @samp{%d} or @samp{%s} (and various other %-sequences
3411 which we have not mentioned). When it sees a control sequence, the
3412 function looks to the second or subsequent arguments and prints the
3413 value of the argument in the location in the string where the control
3414 sequence is located.
3416 In the interactive @code{multiply-by-seven} function, the control string
3417 is @samp{%d}, which requires a number, and the value returned by
3418 evaluating @code{(* 7 5)} is the number 35. Consequently, the number 35
3419 is printed in place of the @samp{%d} and the message is @samp{The result
3422 (Note that when you call the function @code{multiply-by-seven}, the
3423 message is printed without quotes, but when you call @code{message}, the
3424 text is printed in double quotes. This is because the value returned by
3425 @code{message} is what appears in the echo area when you evaluate an
3426 expression whose first element is @code{message}; but when embedded in a
3427 function, @code{message} prints the text as a side effect without
3430 @node Interactive Options
3431 @section Different Options for @code{interactive}
3432 @cindex Options for @code{interactive}
3433 @cindex Interactive options
3435 In the example, @code{multiply-by-seven} used @code{"p"} as the
3436 argument to @code{interactive}. This argument told Emacs to interpret
3437 your typing either @kbd{C-u} followed by a number or @key{META}
3438 followed by a number as a command to pass that number to the function
3439 as its argument. Emacs has more than twenty characters predefined for
3440 use with @code{interactive}. In almost every case, one of these
3441 options will enable you to pass the right information interactively to
3442 a function. (@xref{Interactive Codes, , Code Characters for
3443 @code{interactive}, elisp, The GNU Emacs Lisp Reference Manual}.)
3446 Consider the function @code{zap-to-char}. Its interactive expression
3450 (interactive "p\ncZap to char: ")
3453 The first part of the argument to @code{interactive} is @samp{p}, with
3454 which you are already familiar. This argument tells Emacs to
3455 interpret a `prefix', as a number to be passed to the function. You
3456 can specify a prefix either by typing @kbd{C-u} followed by a number
3457 or by typing @key{META} followed by a number. The prefix is the
3458 number of specified characters. Thus, if your prefix is three and the
3459 specified character is @samp{x}, then you will delete all the text up
3460 to and including the third next @samp{x}. If you do not set a prefix,
3461 then you delete all the text up to and including the specified
3462 character, but no more.
3464 The @samp{c} tells the function the name of the character to which to delete.
3466 More formally, a function with two or more arguments can have
3467 information passed to each argument by adding parts to the string that
3468 follows @code{interactive}. When you do this, the information is
3469 passed to each argument in the same order it is specified in the
3470 @code{interactive} list. In the string, each part is separated from
3471 the next part by a @samp{\n}, which is a newline. For example, you
3472 can follow @samp{p} with a @samp{\n} and an @samp{cZap to char:@: }.
3473 This causes Emacs to pass the value of the prefix argument (if there
3474 is one) and the character.
3476 In this case, the function definition looks like the following, where
3477 @code{arg} and @code{char} are the symbols to which @code{interactive}
3478 binds the prefix argument and the specified character:
3482 (defun @var{name-of-function} (arg char)
3483 "@var{documentation}@dots{}"
3484 (interactive "p\ncZap to char: ")
3485 @var{body-of-function}@dots{})
3490 (The space after the colon in the prompt makes it look better when you
3491 are prompted. @xref{copy-to-buffer, , The Definition of
3492 @code{copy-to-buffer}}, for an example.)
3494 When a function does not take arguments, @code{interactive} does not
3495 require any. Such a function contains the simple expression
3496 @code{(interactive)}. The @code{mark-whole-buffer} function is like
3499 Alternatively, if the special letter-codes are not right for your
3500 application, you can pass your own arguments to @code{interactive} as
3503 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}},
3504 for an example. @xref{Using Interactive, , Using @code{Interactive},
3505 elisp, The GNU Emacs Lisp Reference Manual}, for a more complete
3506 explanation about this technique.
3508 @node Permanent Installation
3509 @section Install Code Permanently
3510 @cindex Install code permanently
3511 @cindex Permanent code installation
3512 @cindex Code installation
3514 When you install a function definition by evaluating it, it will stay
3515 installed until you quit Emacs. The next time you start a new session
3516 of Emacs, the function will not be installed unless you evaluate the
3517 function definition again.
3519 At some point, you may want to have code installed automatically
3520 whenever you start a new session of Emacs. There are several ways of
3525 If you have code that is just for yourself, you can put the code for the
3526 function definition in your @file{.emacs} initialization file. When you
3527 start Emacs, your @file{.emacs} file is automatically evaluated and all
3528 the function definitions within it are installed.
3529 @xref{Emacs Initialization, , Your @file{.emacs} File}.
3532 Alternatively, you can put the function definitions that you want
3533 installed in one or more files of their own and use the @code{load}
3534 function to cause Emacs to evaluate and thereby install each of the
3535 functions in the files.
3536 @xref{Loading Files, , Loading Files}.
3539 Thirdly, if you have code that your whole site will use, it is usual
3540 to put it in a file called @file{site-init.el} that is loaded when
3541 Emacs is built. This makes the code available to everyone who uses
3542 your machine. (See the @file{INSTALL} file that is part of the Emacs
3546 Finally, if you have code that everyone who uses Emacs may want, you
3547 can post it on a computer network or send a copy to the Free Software
3548 Foundation. (When you do this, please license the code and its
3549 documentation under a license that permits other people to run, copy,
3550 study, modify, and redistribute the code and which protects you from
3551 having your work taken from you.) If you send a copy of your code to
3552 the Free Software Foundation, and properly protect yourself and
3553 others, it may be included in the next release of Emacs. In large
3554 part, this is how Emacs has grown over the past years, by donations.
3560 The @code{let} expression is a special form in Lisp that you will need
3561 to use in most function definitions.
3563 @code{let} is used to attach or bind a symbol to a value in such a way
3564 that the Lisp interpreter will not confuse the variable with a
3565 variable of the same name that is not part of the function.
3567 To understand why the @code{let} special form is necessary, consider
3568 the situation in which you own a home that you generally refer to as
3569 `the house', as in the sentence, ``The house needs painting.'' If you
3570 are visiting a friend and your host refers to `the house', he is
3571 likely to be referring to @emph{his} house, not yours, that is, to a
3574 If your friend is referring to his house and you think he is referring
3575 to your house, you may be in for some confusion. The same thing could
3576 happen in Lisp if a variable that is used inside of one function has
3577 the same name as a variable that is used inside of another function,
3578 and the two are not intended to refer to the same value. The
3579 @code{let} special form prevents this kind of confusion.
3582 * Prevent confusion::
3583 * Parts of let Expression::
3584 * Sample let Expression::
3585 * Uninitialized let Variables::
3589 @node Prevent confusion
3590 @unnumberedsubsec @code{let} Prevents Confusion
3593 @cindex @samp{local variable} defined
3594 @cindex @samp{variable, local}, defined
3595 The @code{let} special form prevents confusion. @code{let} creates a
3596 name for a @dfn{local variable} that overshadows any use of the same
3597 name outside the @code{let} expression. This is like understanding
3598 that whenever your host refers to `the house', he means his house, not
3599 yours. (Symbols used in argument lists work the same way.
3600 @xref{defun, , The @code{defun} Macro}.)
3602 Local variables created by a @code{let} expression retain their value
3603 @emph{only} within the @code{let} expression itself (and within
3604 expressions called within the @code{let} expression); the local
3605 variables have no effect outside the @code{let} expression.
3607 Another way to think about @code{let} is that it is like a @code{setq}
3608 that is temporary and local. The values set by @code{let} are
3609 automatically undone when the @code{let} is finished. The setting
3610 only affects expressions that are inside the bounds of the @code{let}
3611 expression. In computer science jargon, we would say ``the binding of
3612 a symbol is visible only in functions called in the @code{let} form;
3613 in Emacs Lisp, scoping is dynamic, not lexical.''
3615 @code{let} can create more than one variable at once. Also,
3616 @code{let} gives each variable it creates an initial value, either a
3617 value specified by you, or @code{nil}. (In the jargon, this is called
3618 `binding the variable to the value'.) After @code{let} has created
3619 and bound the variables, it executes the code in the body of the
3620 @code{let}, and returns the value of the last expression in the body,
3621 as the value of the whole @code{let} expression. (`Execute' is a jargon
3622 term that means to evaluate a list; it comes from the use of the word
3623 meaning `to give practical effect to' (@cite{Oxford English
3624 Dictionary}). Since you evaluate an expression to perform an action,
3625 `execute' has evolved as a synonym to `evaluate'.)
3627 @node Parts of let Expression
3628 @subsection The Parts of a @code{let} Expression
3629 @cindex @code{let} expression, parts of
3630 @cindex Parts of @code{let} expression
3632 @cindex @samp{varlist} defined
3633 A @code{let} expression is a list of three parts. The first part is
3634 the symbol @code{let}. The second part is a list, called a
3635 @dfn{varlist}, each element of which is either a symbol by itself or a
3636 two-element list, the first element of which is a symbol. The third
3637 part of the @code{let} expression is the body of the @code{let}. The
3638 body usually consists of one or more lists.
3641 A template for a @code{let} expression looks like this:
3644 (let @var{varlist} @var{body}@dots{})
3648 The symbols in the varlist are the variables that are given initial
3649 values by the @code{let} special form. Symbols by themselves are given
3650 the initial value of @code{nil}; and each symbol that is the first
3651 element of a two-element list is bound to the value that is returned
3652 when the Lisp interpreter evaluates the second element.
3654 Thus, a varlist might look like this: @code{(thread (needles 3))}. In
3655 this case, in a @code{let} expression, Emacs binds the symbol
3656 @code{thread} to an initial value of @code{nil}, and binds the symbol
3657 @code{needles} to an initial value of 3.
3659 When you write a @code{let} expression, what you do is put the
3660 appropriate expressions in the slots of the @code{let} expression
3663 If the varlist is composed of two-element lists, as is often the case,
3664 the template for the @code{let} expression looks like this:
3668 (let ((@var{variable} @var{value})
3669 (@var{variable} @var{value})
3675 @node Sample let Expression
3676 @subsection Sample @code{let} Expression
3677 @cindex Sample @code{let} expression
3678 @cindex @code{let} expression sample
3680 The following expression creates and gives initial values
3681 to the two variables @code{zebra} and @code{tiger}. The body of the
3682 @code{let} expression is a list which calls the @code{message} function.
3686 (let ((zebra 'stripes)
3688 (message "One kind of animal has %s and another is %s."
3693 Here, the varlist is @code{((zebra 'stripes) (tiger 'fierce))}.
3695 The two variables are @code{zebra} and @code{tiger}. Each variable is
3696 the first element of a two-element list and each value is the second
3697 element of its two-element list. In the varlist, Emacs binds the
3698 variable @code{zebra} to the value @code{stripes}@footnote{According
3699 to Jared Diamond in @cite{Guns, Germs, and Steel}, ``@dots{} zebras
3700 become impossibly dangerous as they grow older'' but the claim here is
3701 that they do not become fierce like a tiger. (1997, W. W. Norton and
3702 Co., ISBN 0-393-03894-2, page 171)}, and binds the
3703 variable @code{tiger} to the value @code{fierce}. In this example,
3704 both values are symbols preceded by a quote. The values could just as
3705 well have been another list or a string. The body of the @code{let}
3706 follows after the list holding the variables. In this example, the
3707 body is a list that uses the @code{message} function to print a string
3711 You may evaluate the example in the usual fashion, by placing the
3712 cursor after the last parenthesis and typing @kbd{C-x C-e}. When you do
3713 this, the following will appear in the echo area:
3716 "One kind of animal has stripes and another is fierce."
3719 As we have seen before, the @code{message} function prints its first
3720 argument, except for @samp{%s}. In this example, the value of the variable
3721 @code{zebra} is printed at the location of the first @samp{%s} and the
3722 value of the variable @code{tiger} is printed at the location of the
3725 @node Uninitialized let Variables
3726 @subsection Uninitialized Variables in a @code{let} Statement
3727 @cindex Uninitialized @code{let} variables
3728 @cindex @code{let} variables uninitialized
3730 If you do not bind the variables in a @code{let} statement to specific
3731 initial values, they will automatically be bound to an initial value of
3732 @code{nil}, as in the following expression:
3741 "Here are %d variables with %s, %s, and %s value."
3742 birch pine fir oak))
3747 Here, the varlist is @code{((birch 3) pine fir (oak 'some))}.
3750 If you evaluate this expression in the usual way, the following will
3751 appear in your echo area:
3754 "Here are 3 variables with nil, nil, and some value."
3758 In this example, Emacs binds the symbol @code{birch} to the number 3,
3759 binds the symbols @code{pine} and @code{fir} to @code{nil}, and binds
3760 the symbol @code{oak} to the value @code{some}.
3762 Note that in the first part of the @code{let}, the variables @code{pine}
3763 and @code{fir} stand alone as atoms that are not surrounded by
3764 parentheses; this is because they are being bound to @code{nil}, the
3765 empty list. But @code{oak} is bound to @code{some} and so is a part of
3766 the list @code{(oak 'some)}. Similarly, @code{birch} is bound to the
3767 number 3 and so is in a list with that number. (Since a number
3768 evaluates to itself, the number does not need to be quoted. Also, the
3769 number is printed in the message using a @samp{%d} rather than a
3770 @samp{%s}.) The four variables as a group are put into a list to
3771 delimit them from the body of the @code{let}.
3774 @section The @code{if} Special Form
3776 @cindex Conditional with @code{if}
3778 A third special form, in addition to @code{defun} and @code{let}, is the
3779 conditional @code{if}. This form is used to instruct the computer to
3780 make decisions. You can write function definitions without using
3781 @code{if}, but it is used often enough, and is important enough, to be
3782 included here. It is used, for example, in the code for the
3783 function @code{beginning-of-buffer}.
3785 The basic idea behind an @code{if}, is that ``@emph{if} a test is true,
3786 @emph{then} an expression is evaluated.'' If the test is not true, the
3787 expression is not evaluated. For example, you might make a decision
3788 such as, ``if it is warm and sunny, then go to the beach!''
3791 * if in more detail::
3792 * type-of-animal in detail:: An example of an @code{if} expression.
3796 @node if in more detail
3797 @unnumberedsubsec @code{if} in more detail
3800 @cindex @samp{if-part} defined
3801 @cindex @samp{then-part} defined
3802 An @code{if} expression written in Lisp does not use the word `then';
3803 the test and the action are the second and third elements of the list
3804 whose first element is @code{if}. Nonetheless, the test part of an
3805 @code{if} expression is often called the @dfn{if-part} and the second
3806 argument is often called the @dfn{then-part}.
3808 Also, when an @code{if} expression is written, the true-or-false-test
3809 is usually written on the same line as the symbol @code{if}, but the
3810 action to carry out if the test is true, the ``then-part'', is written
3811 on the second and subsequent lines. This makes the @code{if}
3812 expression easier to read.
3816 (if @var{true-or-false-test}
3817 @var{action-to-carry-out-if-test-is-true})
3822 The true-or-false-test will be an expression that
3823 is evaluated by the Lisp interpreter.
3825 Here is an example that you can evaluate in the usual manner. The test
3826 is whether the number 5 is greater than the number 4. Since it is, the
3827 message @samp{5 is greater than 4!} will be printed.
3831 (if (> 5 4) ; @r{if-part}
3832 (message "5 is greater than 4!")) ; @r{then-part}
3837 (The function @code{>} tests whether its first argument is greater than
3838 its second argument and returns true if it is.)
3839 @findex > (greater than)
3841 Of course, in actual use, the test in an @code{if} expression will not
3842 be fixed for all time as it is by the expression @code{(> 5 4)}.
3843 Instead, at least one of the variables used in the test will be bound to
3844 a value that is not known ahead of time. (If the value were known ahead
3845 of time, we would not need to run the test!)
3847 For example, the value may be bound to an argument of a function
3848 definition. In the following function definition, the character of the
3849 animal is a value that is passed to the function. If the value bound to
3850 @code{characteristic} is @code{fierce}, then the message, @samp{It's a
3851 tiger!} will be printed; otherwise, @code{nil} will be returned.
3855 (defun type-of-animal (characteristic)
3856 "Print message in echo area depending on CHARACTERISTIC.
3857 If the CHARACTERISTIC is the symbol `fierce',
3858 then warn of a tiger."
3859 (if (equal characteristic 'fierce)
3860 (message "It's a tiger!")))
3866 If you are reading this inside of GNU Emacs, you can evaluate the
3867 function definition in the usual way to install it in Emacs, and then you
3868 can evaluate the following two expressions to see the results:
3872 (type-of-animal 'fierce)
3874 (type-of-animal 'zebra)
3879 @c Following sentences rewritten to prevent overfull hbox.
3881 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
3882 following message printed in the echo area: @code{"It's a tiger!"}; and
3883 when you evaluate @code{(type-of-animal 'zebra)} you will see @code{nil}
3884 printed in the echo area.
3886 @node type-of-animal in detail
3887 @subsection The @code{type-of-animal} Function in Detail
3889 Let's look at the @code{type-of-animal} function in detail.
3891 The function definition for @code{type-of-animal} was written by filling
3892 the slots of two templates, one for a function definition as a whole, and
3893 a second for an @code{if} expression.
3896 The template for every function that is not interactive is:
3900 (defun @var{name-of-function} (@var{argument-list})
3901 "@var{documentation}@dots{}"
3907 The parts of the function that match this template look like this:
3911 (defun type-of-animal (characteristic)
3912 "Print message in echo area depending on CHARACTERISTIC.
3913 If the CHARACTERISTIC is the symbol `fierce',
3914 then warn of a tiger."
3915 @var{body: the} @code{if} @var{expression})
3919 The name of function is @code{type-of-animal}; it is passed the value
3920 of one argument. The argument list is followed by a multi-line
3921 documentation string. The documentation string is included in the
3922 example because it is a good habit to write documentation string for
3923 every function definition. The body of the function definition
3924 consists of the @code{if} expression.
3927 The template for an @code{if} expression looks like this:
3931 (if @var{true-or-false-test}
3932 @var{action-to-carry-out-if-the-test-returns-true})
3937 In the @code{type-of-animal} function, the code for the @code{if}
3942 (if (equal characteristic 'fierce)
3943 (message "It's a tiger!")))
3948 Here, the true-or-false-test is the expression:
3951 (equal characteristic 'fierce)
3955 In Lisp, @code{equal} is a function that determines whether its first
3956 argument is equal to its second argument. The second argument is the
3957 quoted symbol @code{'fierce} and the first argument is the value of the
3958 symbol @code{characteristic}---in other words, the argument passed to
3961 In the first exercise of @code{type-of-animal}, the argument
3962 @code{fierce} is passed to @code{type-of-animal}. Since @code{fierce}
3963 is equal to @code{fierce}, the expression, @code{(equal characteristic
3964 'fierce)}, returns a value of true. When this happens, the @code{if}
3965 evaluates the second argument or then-part of the @code{if}:
3966 @code{(message "It's tiger!")}.
3968 On the other hand, in the second exercise of @code{type-of-animal}, the
3969 argument @code{zebra} is passed to @code{type-of-animal}. @code{zebra}
3970 is not equal to @code{fierce}, so the then-part is not evaluated and
3971 @code{nil} is returned by the @code{if} expression.
3974 @section If--then--else Expressions
3977 An @code{if} expression may have an optional third argument, called
3978 the @dfn{else-part}, for the case when the true-or-false-test returns
3979 false. When this happens, the second argument or then-part of the
3980 overall @code{if} expression is @emph{not} evaluated, but the third or
3981 else-part @emph{is} evaluated. You might think of this as the cloudy
3982 day alternative for the decision ``if it is warm and sunny, then go to
3983 the beach, else read a book!''.
3985 The word ``else'' is not written in the Lisp code; the else-part of an
3986 @code{if} expression comes after the then-part. In the written Lisp, the
3987 else-part is usually written to start on a line of its own and is
3988 indented less than the then-part:
3992 (if @var{true-or-false-test}
3993 @var{action-to-carry-out-if-the-test-returns-true}
3994 @var{action-to-carry-out-if-the-test-returns-false})
3998 For example, the following @code{if} expression prints the message @samp{4
3999 is not greater than 5!} when you evaluate it in the usual way:
4003 (if (> 4 5) ; @r{if-part}
4004 (message "4 falsely greater than 5!") ; @r{then-part}
4005 (message "4 is not greater than 5!")) ; @r{else-part}
4010 Note that the different levels of indentation make it easy to
4011 distinguish the then-part from the else-part. (GNU Emacs has several
4012 commands that automatically indent @code{if} expressions correctly.
4013 @xref{Typing Lists, , GNU Emacs Helps You Type Lists}.)
4015 We can extend the @code{type-of-animal} function to include an
4016 else-part by simply incorporating an additional part to the @code{if}
4020 You can see the consequences of doing this if you evaluate the following
4021 version of the @code{type-of-animal} function definition to install it
4022 and then evaluate the two subsequent expressions to pass different
4023 arguments to the function.
4027 (defun type-of-animal (characteristic) ; @r{Second version.}
4028 "Print message in echo area depending on CHARACTERISTIC.
4029 If the CHARACTERISTIC is the symbol `fierce',
4030 then warn of a tiger;
4031 else say it's not fierce."
4032 (if (equal characteristic 'fierce)
4033 (message "It's a tiger!")
4034 (message "It's not fierce!")))
4041 (type-of-animal 'fierce)
4043 (type-of-animal 'zebra)
4048 @c Following sentence rewritten to prevent overfull hbox.
4050 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
4051 following message printed in the echo area: @code{"It's a tiger!"}; but
4052 when you evaluate @code{(type-of-animal 'zebra)}, you will see
4053 @code{"It's not fierce!"}.
4055 (Of course, if the @var{characteristic} were @code{ferocious}, the
4056 message @code{"It's not fierce!"} would be printed; and it would be
4057 misleading! When you write code, you need to take into account the
4058 possibility that some such argument will be tested by the @code{if}
4059 and write your program accordingly.)
4061 @node Truth & Falsehood
4062 @section Truth and Falsehood in Emacs Lisp
4063 @cindex Truth and falsehood in Emacs Lisp
4064 @cindex Falsehood and truth in Emacs Lisp
4067 There is an important aspect to the truth test in an @code{if}
4068 expression. So far, we have spoken of `true' and `false' as values of
4069 predicates as if they were new kinds of Emacs Lisp objects. In fact,
4070 `false' is just our old friend @code{nil}. Anything else---anything
4073 The expression that tests for truth is interpreted as @dfn{true}
4074 if the result of evaluating it is a value that is not @code{nil}. In
4075 other words, the result of the test is considered true if the value
4076 returned is a number such as 47, a string such as @code{"hello"}, or a
4077 symbol (other than @code{nil}) such as @code{flowers}, or a list (so
4078 long as it is not empty), or even a buffer!
4081 * nil explained:: @code{nil} has two meanings.
4086 @unnumberedsubsec An explanation of @code{nil}
4089 Before illustrating a test for truth, we need an explanation of @code{nil}.
4091 In Emacs Lisp, the symbol @code{nil} has two meanings. First, it means the
4092 empty list. Second, it means false and is the value returned when a
4093 true-or-false-test tests false. @code{nil} can be written as an empty
4094 list, @code{()}, or as @code{nil}. As far as the Lisp interpreter is
4095 concerned, @code{()} and @code{nil} are the same. Humans, however, tend
4096 to use @code{nil} for false and @code{()} for the empty list.
4098 In Emacs Lisp, any value that is not @code{nil}---is not the empty
4099 list---is considered true. This means that if an evaluation returns
4100 something that is not an empty list, an @code{if} expression will test
4101 true. For example, if a number is put in the slot for the test, it
4102 will be evaluated and will return itself, since that is what numbers
4103 do when evaluated. In this conditional, the @code{if} expression will
4104 test true. The expression tests false only when @code{nil}, an empty
4105 list, is returned by evaluating the expression.
4107 You can see this by evaluating the two expressions in the following examples.
4109 In the first example, the number 4 is evaluated as the test in the
4110 @code{if} expression and returns itself; consequently, the then-part
4111 of the expression is evaluated and returned: @samp{true} appears in
4112 the echo area. In the second example, the @code{nil} indicates false;
4113 consequently, the else-part of the expression is evaluated and
4114 returned: @samp{false} appears in the echo area.
4131 Incidentally, if some other useful value is not available for a test that
4132 returns true, then the Lisp interpreter will return the symbol @code{t}
4133 for true. For example, the expression @code{(> 5 4)} returns @code{t}
4134 when evaluated, as you can see by evaluating it in the usual way:
4142 On the other hand, this function returns @code{nil} if the test is false.
4148 @node save-excursion
4149 @section @code{save-excursion}
4150 @findex save-excursion
4151 @cindex Region, what it is
4152 @cindex Preserving point, mark, and buffer
4153 @cindex Point, mark, buffer preservation
4157 The @code{save-excursion} function is the third and final special form
4158 that we will discuss in this chapter.
4160 In Emacs Lisp programs used for editing, the @code{save-excursion}
4161 function is very common. It saves the location of point and mark,
4162 executes the body of the function, and then restores point and mark to
4163 their previous positions if their locations were changed. Its primary
4164 purpose is to keep the user from being surprised and disturbed by
4165 unexpected movement of point or mark.
4168 * Point and mark:: A review of various locations.
4169 * Template for save-excursion::
4173 @node Point and mark
4174 @unnumberedsubsec Point and Mark
4177 Before discussing @code{save-excursion}, however, it may be useful
4178 first to review what point and mark are in GNU Emacs. @dfn{Point} is
4179 the current location of the cursor. Wherever the cursor
4180 is, that is point. More precisely, on terminals where the cursor
4181 appears to be on top of a character, point is immediately before the
4182 character. In Emacs Lisp, point is an integer. The first character in
4183 a buffer is number one, the second is number two, and so on. The
4184 function @code{point} returns the current position of the cursor as a
4185 number. Each buffer has its own value for point.
4187 The @dfn{mark} is another position in the buffer; its value can be set
4188 with a command such as @kbd{C-@key{SPC}} (@code{set-mark-command}). If
4189 a mark has been set, you can use the command @kbd{C-x C-x}
4190 (@code{exchange-point-and-mark}) to cause the cursor to jump to the mark
4191 and set the mark to be the previous position of point. In addition, if
4192 you set another mark, the position of the previous mark is saved in the
4193 mark ring. Many mark positions can be saved this way. You can jump the
4194 cursor to a saved mark by typing @kbd{C-u C-@key{SPC}} one or more
4197 The part of the buffer between point and mark is called @dfn{the
4198 region}. Numerous commands work on the region, including
4199 @code{center-region}, @code{count-lines-region}, @code{kill-region}, and
4200 @code{print-region}.
4202 The @code{save-excursion} special form saves the locations of point and
4203 mark and restores those positions after the code within the body of the
4204 special form is evaluated by the Lisp interpreter. Thus, if point were
4205 in the beginning of a piece of text and some code moved point to the end
4206 of the buffer, the @code{save-excursion} would put point back to where
4207 it was before, after the expressions in the body of the function were
4210 In Emacs, a function frequently moves point as part of its internal
4211 workings even though a user would not expect this. For example,
4212 @code{count-lines-region} moves point. To prevent the user from being
4213 bothered by jumps that are both unexpected and (from the user's point of
4214 view) unnecessary, @code{save-excursion} is often used to keep point and
4215 mark in the location expected by the user. The use of
4216 @code{save-excursion} is good housekeeping.
4218 To make sure the house stays clean, @code{save-excursion} restores the
4219 values of point and mark even if something goes wrong in the code inside
4220 of it (or, to be more precise and to use the proper jargon, ``in case of
4221 abnormal exit''). This feature is very helpful.
4223 In addition to recording the values of point and mark,
4224 @code{save-excursion} keeps track of the current buffer, and restores
4225 it, too. This means you can write code that will change the buffer and
4226 have @code{save-excursion} switch you back to the original buffer.
4227 This is how @code{save-excursion} is used in @code{append-to-buffer}.
4228 (@xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
4230 @node Template for save-excursion
4231 @subsection Template for a @code{save-excursion} Expression
4234 The template for code using @code{save-excursion} is simple:
4244 The body of the function is one or more expressions that will be
4245 evaluated in sequence by the Lisp interpreter. If there is more than
4246 one expression in the body, the value of the last one will be returned
4247 as the value of the @code{save-excursion} function. The other
4248 expressions in the body are evaluated only for their side effects; and
4249 @code{save-excursion} itself is used only for its side effect (which
4250 is restoring the positions of point and mark).
4253 In more detail, the template for a @code{save-excursion} expression
4259 @var{first-expression-in-body}
4260 @var{second-expression-in-body}
4261 @var{third-expression-in-body}
4263 @var{last-expression-in-body})
4268 An expression, of course, may be a symbol on its own or a list.
4270 In Emacs Lisp code, a @code{save-excursion} expression often occurs
4271 within the body of a @code{let} expression. It looks like this:
4284 In the last few chapters we have introduced a macro and a fair number
4285 of functions and special forms. Here they are described in brief,
4286 along with a few similar functions that have not been mentioned yet.
4289 @item eval-last-sexp
4290 Evaluate the last symbolic expression before the current location of
4291 point. The value is printed in the echo area unless the function is
4292 invoked with an argument; in that case, the output is printed in the
4293 current buffer. This command is normally bound to @kbd{C-x C-e}.
4296 Define function. This macro has up to five parts: the name, a
4297 template for the arguments that will be passed to the function,
4298 documentation, an optional interactive declaration, and the body of
4302 For example, in an early version of Emacs, the function definition was
4303 as follows. (It is slightly more complex now that it seeks the first
4304 non-whitespace character rather than the first visible character.)
4308 (defun back-to-indentation ()
4309 "Move point to first visible character on line."
4311 (beginning-of-line 1)
4312 (skip-chars-forward " \t"))
4319 (defun backward-to-indentation (&optional arg)
4320 "Move backward ARG lines and position at first nonblank character."
4322 (forward-line (- (or arg 1)))
4323 (skip-chars-forward " \t"))
4325 (defun back-to-indentation ()
4326 "Move point to the first non-whitespace character on this line."
4328 (beginning-of-line 1)
4329 (skip-syntax-forward " " (line-end-position))
4330 ;; Move back over chars that have whitespace syntax but have the p flag.
4331 (backward-prefix-chars))
4335 Declare to the interpreter that the function can be used
4336 interactively. This special form may be followed by a string with one
4337 or more parts that pass the information to the arguments of the
4338 function, in sequence. These parts may also tell the interpreter to
4339 prompt for information. Parts of the string are separated by
4340 newlines, @samp{\n}.
4343 Common code characters are:
4347 The name of an existing buffer.
4350 The name of an existing file.
4353 The numeric prefix argument. (Note that this `p' is lower case.)
4356 Point and the mark, as two numeric arguments, smallest first. This
4357 is the only code letter that specifies two successive arguments
4361 @xref{Interactive Codes, , Code Characters for @samp{interactive},
4362 elisp, The GNU Emacs Lisp Reference Manual}, for a complete list of
4366 Declare that a list of variables is for use within the body of the
4367 @code{let} and give them an initial value, either @code{nil} or a
4368 specified value; then evaluate the rest of the expressions in the body
4369 of the @code{let} and return the value of the last one. Inside the
4370 body of the @code{let}, the Lisp interpreter does not see the values of
4371 the variables of the same names that are bound outside of the
4379 (let ((foo (buffer-name))
4380 (bar (buffer-size)))
4382 "This buffer is %s and has %d characters."
4387 @item save-excursion
4388 Record the values of point and mark and the current buffer before
4389 evaluating the body of this special form. Restore the values of point
4390 and mark and buffer afterward.
4397 (message "We are %d characters into this buffer."
4400 (goto-char (point-min)) (point))))
4405 Evaluate the first argument to the function; if it is true, evaluate
4406 the second argument; else evaluate the third argument, if there is one.
4408 The @code{if} special form is called a @dfn{conditional}. There are
4409 other conditionals in Emacs Lisp, but @code{if} is perhaps the most
4417 (if (= 22 emacs-major-version)
4418 (message "This is version 22 Emacs")
4419 (message "This is not version 22 Emacs"))
4428 The @code{<} function tests whether its first argument is smaller than
4429 its second argument. A corresponding function, @code{>}, tests whether
4430 the first argument is greater than the second. Likewise, @code{<=}
4431 tests whether the first argument is less than or equal to the second and
4432 @code{>=} tests whether the first argument is greater than or equal to
4433 the second. In all cases, both arguments must be numbers or markers
4434 (markers indicate positions in buffers).
4438 The @code{=} function tests whether two arguments, both numbers or
4444 Test whether two objects are the same. @code{equal} uses one meaning
4445 of the word `same' and @code{eq} uses another: @code{equal} returns
4446 true if the two objects have a similar structure and contents, such as
4447 two copies of the same book. On the other hand, @code{eq}, returns
4448 true if both arguments are actually the same object.
4457 The @code{string-lessp} function tests whether its first argument is
4458 smaller than the second argument. A shorter, alternative name for the
4459 same function (a @code{defalias}) is @code{string<}.
4461 The arguments to @code{string-lessp} must be strings or symbols; the
4462 ordering is lexicographic, so case is significant. The print names of
4463 symbols are used instead of the symbols themselves.
4465 @cindex @samp{empty string} defined
4466 An empty string, @samp{""}, a string with no characters in it, is
4467 smaller than any string of characters.
4469 @code{string-equal} provides the corresponding test for equality. Its
4470 shorter, alternative name is @code{string=}. There are no string test
4471 functions that correspond to @var{>}, @code{>=}, or @code{<=}.
4474 Print a message in the echo area. The first argument is a string that
4475 can contain @samp{%s}, @samp{%d}, or @samp{%c} to print the value of
4476 arguments that follow the string. The argument used by @samp{%s} must
4477 be a string or a symbol; the argument used by @samp{%d} must be a
4478 number. The argument used by @samp{%c} must be an @sc{ascii} code
4479 number; it will be printed as the character with that @sc{ascii} code.
4480 (Various other %-sequences have not been mentioned.)
4484 The @code{setq} function sets the value of its first argument to the
4485 value of the second argument. The first argument is automatically
4486 quoted by @code{setq}. It does the same for succeeding pairs of
4487 arguments. Another function, @code{set}, takes only two arguments and
4488 evaluates both of them before setting the value returned by its first
4489 argument to the value returned by its second argument.
4492 Without an argument, return the name of the buffer, as a string.
4494 @item buffer-file-name
4495 Without an argument, return the name of the file the buffer is
4498 @item current-buffer
4499 Return the buffer in which Emacs is active; it may not be
4500 the buffer that is visible on the screen.
4503 Return the most recently selected buffer (other than the buffer passed
4504 to @code{other-buffer} as an argument and other than the current
4507 @item switch-to-buffer
4508 Select a buffer for Emacs to be active in and display it in the current
4509 window so users can look at it. Usually bound to @kbd{C-x b}.
4512 Switch Emacs's attention to a buffer on which programs will run. Don't
4513 alter what the window is showing.
4516 Return the number of characters in the current buffer.
4519 Return the value of the current position of the cursor, as an
4520 integer counting the number of characters from the beginning of the
4524 Return the minimum permissible value of point in
4525 the current buffer. This is 1, unless narrowing is in effect.
4528 Return the value of the maximum permissible value of point in the
4529 current buffer. This is the end of the buffer, unless narrowing is in
4534 @node defun Exercises
4539 Write a non-interactive function that doubles the value of its
4540 argument, a number. Make that function interactive.
4543 Write a function that tests whether the current value of
4544 @code{fill-column} is greater than the argument passed to the function,
4545 and if so, prints an appropriate message.
4548 @node Buffer Walk Through
4549 @chapter A Few Buffer--Related Functions
4551 In this chapter we study in detail several of the functions used in GNU
4552 Emacs. This is called a ``walk-through''. These functions are used as
4553 examples of Lisp code, but are not imaginary examples; with the
4554 exception of the first, simplified function definition, these functions
4555 show the actual code used in GNU Emacs. You can learn a great deal from
4556 these definitions. The functions described here are all related to
4557 buffers. Later, we will study other functions.
4560 * Finding More:: How to find more information.
4561 * simplified-beginning-of-buffer:: Shows @code{goto-char},
4562 @code{point-min}, and @code{push-mark}.
4563 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
4564 * append-to-buffer:: Uses @code{save-excursion} and
4565 @code{insert-buffer-substring}.
4566 * Buffer Related Review:: Review.
4567 * Buffer Exercises::
4571 @section Finding More Information
4573 @findex describe-function, @r{introduced}
4574 @cindex Find function documentation
4575 In this walk-through, I will describe each new function as we come to
4576 it, sometimes in detail and sometimes briefly. If you are interested,
4577 you can get the full documentation of any Emacs Lisp function at any
4578 time by typing @kbd{C-h f} and then the name of the function (and then
4579 @key{RET}). Similarly, you can get the full documentation for a
4580 variable by typing @kbd{C-h v} and then the name of the variable (and
4583 @cindex Find source of function
4584 @c In version 22, tells location both of C and of Emacs Lisp
4585 Also, @code{describe-function} will tell you the location of the
4586 function definition.
4588 Put point into the name of the file that contains the function and
4589 press the @key{RET} key. In this case, @key{RET} means
4590 @code{push-button} rather than `return' or `enter'. Emacs will take
4591 you directly to the function definition.
4596 If you move point over the file name and press
4597 the @key{RET} key, which in this case means @code{help-follow} rather
4598 than `return' or `enter', Emacs will take you directly to the function
4602 More generally, if you want to see a function in its original source
4603 file, you can use the @code{find-tag} function to jump to it.
4604 @code{find-tag} works with a wide variety of languages, not just
4605 Lisp, and C, and it works with non-programming text as well. For
4606 example, @code{find-tag} will jump to the various nodes in the
4607 Texinfo source file of this document.
4608 The @code{find-tag} function depends on `tags tables' that record
4609 the locations of the functions, variables, and other items to which
4610 @code{find-tag} jumps.
4612 To use the @code{find-tag} command, type @kbd{M-.} (i.e., press the
4613 period key while holding down the @key{META} key, or else type the
4614 @key{ESC} key and then type the period key), and then, at the prompt,
4615 type in the name of the function whose source code you want to see,
4616 such as @code{mark-whole-buffer}, and then type @key{RET}. Emacs will
4617 switch buffers and display the source code for the function on your
4618 screen. To switch back to your current buffer, type @kbd{C-x b
4619 @key{RET}}. (On some keyboards, the @key{META} key is labeled
4622 @c !!! 22.1.1 tags table location in this paragraph
4623 @cindex TAGS table, specifying
4625 Depending on how the initial default values of your copy of Emacs are
4626 set, you may also need to specify the location of your `tags table',
4627 which is a file called @file{TAGS}. For example, if you are
4628 interested in Emacs sources, the tags table you will most likely want,
4629 if it has already been created for you, will be in a subdirectory of
4630 the @file{/usr/local/share/emacs/} directory; thus you would use the
4631 @code{M-x visit-tags-table} command and specify a pathname such as
4632 @file{/usr/local/share/emacs/22.1.1/lisp/TAGS}. If the tags table
4633 has not already been created, you will have to create it yourself. It
4634 will be in a file such as @file{/usr/local/src/emacs/src/TAGS}.
4637 To create a @file{TAGS} file in a specific directory, switch to that
4638 directory in Emacs using @kbd{M-x cd} command, or list the directory
4639 with @kbd{C-x d} (@code{dired}). Then run the compile command, with
4640 @w{@code{etags *.el}} as the command to execute:
4643 M-x compile RET etags *.el RET
4646 For more information, see @ref{etags, , Create Your Own @file{TAGS} File}.
4648 After you become more familiar with Emacs Lisp, you will find that you will
4649 frequently use @code{find-tag} to navigate your way around source code;
4650 and you will create your own @file{TAGS} tables.
4652 @cindex Library, as term for `file'
4653 Incidentally, the files that contain Lisp code are conventionally
4654 called @dfn{libraries}. The metaphor is derived from that of a
4655 specialized library, such as a law library or an engineering library,
4656 rather than a general library. Each library, or file, contains
4657 functions that relate to a particular topic or activity, such as
4658 @file{abbrev.el} for handling abbreviations and other typing
4659 shortcuts, and @file{help.el} for on-line help. (Sometimes several
4660 libraries provide code for a single activity, as the various
4661 @file{rmail@dots{}} files provide code for reading electronic mail.)
4662 In @cite{The GNU Emacs Manual}, you will see sentences such as ``The
4663 @kbd{C-h p} command lets you search the standard Emacs Lisp libraries
4664 by topic keywords.''
4666 @node simplified-beginning-of-buffer
4667 @section A Simplified @code{beginning-of-buffer} Definition
4668 @findex simplified-beginning-of-buffer
4670 The @code{beginning-of-buffer} command is a good function to start with
4671 since you are likely to be familiar with it and it is easy to
4672 understand. Used as an interactive command, @code{beginning-of-buffer}
4673 moves the cursor to the beginning of the buffer, leaving the mark at the
4674 previous position. It is generally bound to @kbd{M-<}.
4676 In this section, we will discuss a shortened version of the function
4677 that shows how it is most frequently used. This shortened function
4678 works as written, but it does not contain the code for a complex option.
4679 In another section, we will describe the entire function.
4680 (@xref{beginning-of-buffer, , Complete Definition of
4681 @code{beginning-of-buffer}}.)
4683 Before looking at the code, let's consider what the function
4684 definition has to contain: it must include an expression that makes
4685 the function interactive so it can be called by typing @kbd{M-x
4686 beginning-of-buffer} or by typing a keychord such as @kbd{M-<}; it
4687 must include code to leave a mark at the original position in the
4688 buffer; and it must include code to move the cursor to the beginning
4692 Here is the complete text of the shortened version of the function:
4696 (defun simplified-beginning-of-buffer ()
4697 "Move point to the beginning of the buffer;
4698 leave mark at previous position."
4701 (goto-char (point-min)))
4705 Like all function definitions, this definition has five parts following
4706 the macro @code{defun}:
4710 The name: in this example, @code{simplified-beginning-of-buffer}.
4713 A list of the arguments: in this example, an empty list, @code{()},
4716 The documentation string.
4719 The interactive expression.
4726 In this function definition, the argument list is empty; this means that
4727 this function does not require any arguments. (When we look at the
4728 definition for the complete function, we will see that it may be passed
4729 an optional argument.)
4731 The interactive expression tells Emacs that the function is intended to
4732 be used interactively. In this example, @code{interactive} does not have
4733 an argument because @code{simplified-beginning-of-buffer} does not
4737 The body of the function consists of the two lines:
4742 (goto-char (point-min))
4746 The first of these lines is the expression, @code{(push-mark)}. When
4747 this expression is evaluated by the Lisp interpreter, it sets a mark at
4748 the current position of the cursor, wherever that may be. The position
4749 of this mark is saved in the mark ring.
4751 The next line is @code{(goto-char (point-min))}. This expression
4752 jumps the cursor to the minimum point in the buffer, that is, to the
4753 beginning of the buffer (or to the beginning of the accessible portion
4754 of the buffer if it is narrowed. @xref{Narrowing & Widening, ,
4755 Narrowing and Widening}.)
4757 The @code{push-mark} command sets a mark at the place where the cursor
4758 was located before it was moved to the beginning of the buffer by the
4759 @code{(goto-char (point-min))} expression. Consequently, you can, if
4760 you wish, go back to where you were originally by typing @kbd{C-x C-x}.
4762 That is all there is to the function definition!
4764 @findex describe-function
4765 When you are reading code such as this and come upon an unfamiliar
4766 function, such as @code{goto-char}, you can find out what it does by
4767 using the @code{describe-function} command. To use this command, type
4768 @kbd{C-h f} and then type in the name of the function and press
4769 @key{RET}. The @code{describe-function} command will print the
4770 function's documentation string in a @file{*Help*} window. For
4771 example, the documentation for @code{goto-char} is:
4775 Set point to POSITION, a number or marker.
4776 Beginning of buffer is position (point-min), end is (point-max).
4781 The function's one argument is the desired position.
4784 (The prompt for @code{describe-function} will offer you the symbol
4785 under or preceding the cursor, so you can save typing by positioning
4786 the cursor right over or after the function and then typing @kbd{C-h f
4789 The @code{end-of-buffer} function definition is written in the same way as
4790 the @code{beginning-of-buffer} definition except that the body of the
4791 function contains the expression @code{(goto-char (point-max))} in place
4792 of @code{(goto-char (point-min))}.
4794 @node mark-whole-buffer
4795 @section The Definition of @code{mark-whole-buffer}
4796 @findex mark-whole-buffer
4798 The @code{mark-whole-buffer} function is no harder to understand than the
4799 @code{simplified-beginning-of-buffer} function. In this case, however,
4800 we will look at the complete function, not a shortened version.
4802 The @code{mark-whole-buffer} function is not as commonly used as the
4803 @code{beginning-of-buffer} function, but is useful nonetheless: it
4804 marks a whole buffer as a region by putting point at the beginning and
4805 a mark at the end of the buffer. It is generally bound to @kbd{C-x
4809 * mark-whole-buffer overview::
4810 * Body of mark-whole-buffer:: Only three lines of code.
4814 @node mark-whole-buffer overview
4815 @unnumberedsubsec An overview of @code{mark-whole-buffer}
4819 In GNU Emacs 22, the code for the complete function looks like this:
4823 (defun mark-whole-buffer ()
4824 "Put point at beginning and mark at end of buffer.
4825 You probably should not use this function in Lisp programs;
4826 it is usually a mistake for a Lisp function to use any subroutine
4827 that uses or sets the mark."
4830 (push-mark (point-max) nil t)
4831 (goto-char (point-min)))
4836 Like all other functions, the @code{mark-whole-buffer} function fits
4837 into the template for a function definition. The template looks like
4842 (defun @var{name-of-function} (@var{argument-list})
4843 "@var{documentation}@dots{}"
4844 (@var{interactive-expression}@dots{})
4849 Here is how the function works: the name of the function is
4850 @code{mark-whole-buffer}; it is followed by an empty argument list,
4851 @samp{()}, which means that the function does not require arguments.
4852 The documentation comes next.
4854 The next line is an @code{(interactive)} expression that tells Emacs
4855 that the function will be used interactively. These details are similar
4856 to the @code{simplified-beginning-of-buffer} function described in the
4860 @node Body of mark-whole-buffer
4861 @subsection Body of @code{mark-whole-buffer}
4863 The body of the @code{mark-whole-buffer} function consists of three
4870 (push-mark (point-max) nil t)
4871 (goto-char (point-min))
4875 The first of these lines is the expression, @code{(push-mark (point))}.
4877 This line does exactly the same job as the first line of the body of
4878 the @code{simplified-beginning-of-buffer} function, which is written
4879 @code{(push-mark)}. In both cases, the Lisp interpreter sets a mark
4880 at the current position of the cursor.
4882 I don't know why the expression in @code{mark-whole-buffer} is written
4883 @code{(push-mark (point))} and the expression in
4884 @code{beginning-of-buffer} is written @code{(push-mark)}. Perhaps
4885 whoever wrote the code did not know that the arguments for
4886 @code{push-mark} are optional and that if @code{push-mark} is not
4887 passed an argument, the function automatically sets mark at the
4888 location of point by default. Or perhaps the expression was written
4889 so as to parallel the structure of the next line. In any case, the
4890 line causes Emacs to determine the position of point and set a mark
4893 In earlier versions of GNU Emacs, the next line of
4894 @code{mark-whole-buffer} was @code{(push-mark (point-max))}. This
4895 expression sets a mark at the point in the buffer that has the highest
4896 number. This will be the end of the buffer (or, if the buffer is
4897 narrowed, the end of the accessible portion of the buffer.
4898 @xref{Narrowing & Widening, , Narrowing and Widening}, for more about
4899 narrowing.) After this mark has been set, the previous mark, the one
4900 set at point, is no longer set, but Emacs remembers its position, just
4901 as all other recent marks are always remembered. This means that you
4902 can, if you wish, go back to that position by typing @kbd{C-u
4906 In GNU Emacs 22, the @code{(point-max)} is slightly more complicated.
4910 (push-mark (point-max) nil t)
4914 The expression works nearly the same as before. It sets a mark at the
4915 highest numbered place in the buffer that it can. However, in this
4916 version, @code{push-mark} has two additional arguments. The second
4917 argument to @code{push-mark} is @code{nil}. This tells the function
4918 it @emph{should} display a message that says `Mark set' when it pushes
4919 the mark. The third argument is @code{t}. This tells
4920 @code{push-mark} to activate the mark when Transient Mark mode is
4921 turned on. Transient Mark mode highlights the currently active
4922 region. It is often turned off.
4924 Finally, the last line of the function is @code{(goto-char
4925 (point-min)))}. This is written exactly the same way as it is written
4926 in @code{beginning-of-buffer}. The expression moves the cursor to
4927 the minimum point in the buffer, that is, to the beginning of the buffer
4928 (or to the beginning of the accessible portion of the buffer). As a
4929 result of this, point is placed at the beginning of the buffer and mark
4930 is set at the end of the buffer. The whole buffer is, therefore, the
4933 @node append-to-buffer
4934 @section The Definition of @code{append-to-buffer}
4935 @findex append-to-buffer
4937 The @code{append-to-buffer} command is more complex than the
4938 @code{mark-whole-buffer} command. What it does is copy the region
4939 (that is, the part of the buffer between point and mark) from the
4940 current buffer to a specified buffer.
4943 * append-to-buffer overview::
4944 * append interactive:: A two part interactive expression.
4945 * append-to-buffer body:: Incorporates a @code{let} expression.
4946 * append save-excursion:: How the @code{save-excursion} works.
4950 @node append-to-buffer overview
4951 @unnumberedsubsec An Overview of @code{append-to-buffer}
4954 @findex insert-buffer-substring
4955 The @code{append-to-buffer} command uses the
4956 @code{insert-buffer-substring} function to copy the region.
4957 @code{insert-buffer-substring} is described by its name: it takes a
4958 string of characters from part of a buffer, a ``substring'', and
4959 inserts them into another buffer.
4961 Most of @code{append-to-buffer} is
4962 concerned with setting up the conditions for
4963 @code{insert-buffer-substring} to work: the code must specify both the
4964 buffer to which the text will go, the window it comes from and goes
4965 to, and the region that will be copied.
4968 Here is the complete text of the function:
4972 (defun append-to-buffer (buffer start end)
4973 "Append to specified buffer the text of the region.
4974 It is inserted into that buffer before its point.
4978 When calling from a program, give three arguments:
4979 BUFFER (or buffer name), START and END.
4980 START and END specify the portion of the current buffer to be copied."
4982 (list (read-buffer "Append to buffer: " (other-buffer
4983 (current-buffer) t))
4984 (region-beginning) (region-end)))
4987 (let ((oldbuf (current-buffer)))
4989 (let* ((append-to (get-buffer-create buffer))
4990 (windows (get-buffer-window-list append-to t t))
4992 (set-buffer append-to)
4993 (setq point (point))
4994 (barf-if-buffer-read-only)
4995 (insert-buffer-substring oldbuf start end)
4996 (dolist (window windows)
4997 (when (= (window-point window) point)
4998 (set-window-point window (point))))))))
5002 The function can be understood by looking at it as a series of
5003 filled-in templates.
5005 The outermost template is for the function definition. In this
5006 function, it looks like this (with several slots filled in):
5010 (defun append-to-buffer (buffer start end)
5011 "@var{documentation}@dots{}"
5012 (interactive @dots{})
5017 The first line of the function includes its name and three arguments.
5018 The arguments are the @code{buffer} to which the text will be copied, and
5019 the @code{start} and @code{end} of the region in the current buffer that
5022 The next part of the function is the documentation, which is clear and
5023 complete. As is conventional, the three arguments are written in
5024 upper case so you will notice them easily. Even better, they are
5025 described in the same order as in the argument list.
5027 Note that the documentation distinguishes between a buffer and its
5028 name. (The function can handle either.)
5030 @node append interactive
5031 @subsection The @code{append-to-buffer} Interactive Expression
5033 Since the @code{append-to-buffer} function will be used interactively,
5034 the function must have an @code{interactive} expression. (For a
5035 review of @code{interactive}, see @ref{Interactive, , Making a
5036 Function Interactive}.) The expression reads as follows:
5042 "Append to buffer: "
5043 (other-buffer (current-buffer) t))
5050 This expression is not one with letters standing for parts, as
5051 described earlier. Instead, it starts a list with these parts:
5053 The first part of the list is an expression to read the name of a
5054 buffer and return it as a string. That is @code{read-buffer}. The
5055 function requires a prompt as its first argument, @samp{"Append to
5056 buffer: "}. Its second argument tells the command what value to
5057 provide if you don't specify anything.
5059 In this case that second argument is an expression containing the
5060 function @code{other-buffer}, an exception, and a @samp{t}, standing
5063 The first argument to @code{other-buffer}, the exception, is yet
5064 another function, @code{current-buffer}. That is not going to be
5065 returned. The second argument is the symbol for true, @code{t}. that
5066 tells @code{other-buffer} that it may show visible buffers (except in
5067 this case, it will not show the current buffer, which makes sense).
5070 The expression looks like this:
5073 (other-buffer (current-buffer) t)
5076 The second and third arguments to the @code{list} expression are
5077 @code{(region-beginning)} and @code{(region-end)}. These two
5078 functions specify the beginning and end of the text to be appended.
5081 Originally, the command used the letters @samp{B} and @samp{r}.
5082 The whole @code{interactive} expression looked like this:
5085 (interactive "BAppend to buffer:@: \nr")
5089 But when that was done, the default value of the buffer switched to
5090 was invisible. That was not wanted.
5092 (The prompt was separated from the second argument with a newline,
5093 @samp{\n}. It was followed by an @samp{r} that told Emacs to bind the
5094 two arguments that follow the symbol @code{buffer} in the function's
5095 argument list (that is, @code{start} and @code{end}) to the values of
5096 point and mark. That argument worked fine.)
5098 @node append-to-buffer body
5099 @subsection The Body of @code{append-to-buffer}
5102 in GNU Emacs 22 in /usr/local/src/emacs/lisp/simple.el
5104 (defun append-to-buffer (buffer start end)
5105 "Append to specified buffer the text of the region.
5106 It is inserted into that buffer before its point.
5108 When calling from a program, give three arguments:
5109 BUFFER (or buffer name), START and END.
5110 START and END specify the portion of the current buffer to be copied."
5112 (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t))
5113 (region-beginning) (region-end)))
5114 (let ((oldbuf (current-buffer)))
5116 (let* ((append-to (get-buffer-create buffer))
5117 (windows (get-buffer-window-list append-to t t))
5119 (set-buffer append-to)
5120 (setq point (point))
5121 (barf-if-buffer-read-only)
5122 (insert-buffer-substring oldbuf start end)
5123 (dolist (window windows)
5124 (when (= (window-point window) point)
5125 (set-window-point window (point))))))))
5128 The body of the @code{append-to-buffer} function begins with @code{let}.
5130 As we have seen before (@pxref{let, , @code{let}}), the purpose of a
5131 @code{let} expression is to create and give initial values to one or
5132 more variables that will only be used within the body of the
5133 @code{let}. This means that such a variable will not be confused with
5134 any variable of the same name outside the @code{let} expression.
5136 We can see how the @code{let} expression fits into the function as a
5137 whole by showing a template for @code{append-to-buffer} with the
5138 @code{let} expression in outline:
5142 (defun append-to-buffer (buffer start end)
5143 "@var{documentation}@dots{}"
5144 (interactive @dots{})
5145 (let ((@var{variable} @var{value}))
5150 The @code{let} expression has three elements:
5154 The symbol @code{let};
5157 A varlist containing, in this case, a single two-element list,
5158 @code{(@var{variable} @var{value})};
5161 The body of the @code{let} expression.
5165 In the @code{append-to-buffer} function, the varlist looks like this:
5168 (oldbuf (current-buffer))
5172 In this part of the @code{let} expression, the one variable,
5173 @code{oldbuf}, is bound to the value returned by the
5174 @code{(current-buffer)} expression. The variable, @code{oldbuf}, is
5175 used to keep track of the buffer in which you are working and from
5176 which you will copy.
5178 The element or elements of a varlist are surrounded by a set of
5179 parentheses so the Lisp interpreter can distinguish the varlist from
5180 the body of the @code{let}. As a consequence, the two-element list
5181 within the varlist is surrounded by a circumscribing set of parentheses.
5182 The line looks like this:
5186 (let ((oldbuf (current-buffer)))
5192 The two parentheses before @code{oldbuf} might surprise you if you did
5193 not realize that the first parenthesis before @code{oldbuf} marks the
5194 boundary of the varlist and the second parenthesis marks the beginning
5195 of the two-element list, @code{(oldbuf (current-buffer))}.
5197 @node append save-excursion
5198 @subsection @code{save-excursion} in @code{append-to-buffer}
5200 The body of the @code{let} expression in @code{append-to-buffer}
5201 consists of a @code{save-excursion} expression.
5203 The @code{save-excursion} function saves the locations of point and
5204 mark, and restores them to those positions after the expressions in the
5205 body of the @code{save-excursion} complete execution. In addition,
5206 @code{save-excursion} keeps track of the original buffer, and
5207 restores it. This is how @code{save-excursion} is used in
5208 @code{append-to-buffer}.
5211 @cindex Indentation for formatting
5212 @cindex Formatting convention
5213 Incidentally, it is worth noting here that a Lisp function is normally
5214 formatted so that everything that is enclosed in a multi-line spread is
5215 indented more to the right than the first symbol. In this function
5216 definition, the @code{let} is indented more than the @code{defun}, and
5217 the @code{save-excursion} is indented more than the @code{let}, like
5233 This formatting convention makes it easy to see that the lines in
5234 the body of the @code{save-excursion} are enclosed by the parentheses
5235 associated with @code{save-excursion}, just as the
5236 @code{save-excursion} itself is enclosed by the parentheses associated
5237 with the @code{let}:
5241 (let ((oldbuf (current-buffer)))
5244 (set-buffer @dots{})
5245 (insert-buffer-substring oldbuf start end)
5251 The use of the @code{save-excursion} function can be viewed as a process
5252 of filling in the slots of a template:
5257 @var{first-expression-in-body}
5258 @var{second-expression-in-body}
5260 @var{last-expression-in-body})
5266 In this function, the body of the @code{save-excursion} contains only
5267 one expression, the @code{let*} expression. You know about a
5268 @code{let} function. The @code{let*} function is different. It has a
5269 @samp{*} in its name. It enables Emacs to set each variable in its
5270 varlist in sequence, one after another.
5272 Its critical feature is that variables later in the varlist can make
5273 use of the values to which Emacs set variables earlier in the varlist.
5274 @xref{fwd-para let, , The @code{let*} expression}.
5276 We will skip functions like @code{let*} and focus on two: the
5277 @code{set-buffer} function and the @code{insert-buffer-substring}
5281 In the old days, the @code{set-buffer} expression was simply
5284 (set-buffer (get-buffer-create buffer))
5292 (set-buffer append-to)
5296 @code{append-to} is bound to @code{(get-buffer-create buffer)} earlier
5297 on in the @code{let*} expression. That extra binding would not be
5298 necessary except for that @code{append-to} is used later in the
5299 varlist as an argument to @code{get-buffer-window-list}.
5304 (let ((oldbuf (current-buffer)))
5306 (let* ((append-to (get-buffer-create buffer))
5307 (windows (get-buffer-window-list append-to t t))
5309 (set-buffer append-to)
5310 (setq point (point))
5311 (barf-if-buffer-read-only)
5312 (insert-buffer-substring oldbuf start end)
5313 (dolist (window windows)
5314 (when (= (window-point window) point)
5315 (set-window-point window (point))))))))
5318 The @code{append-to-buffer} function definition inserts text from the
5319 buffer in which you are currently to a named buffer. It happens that
5320 @code{insert-buffer-substring} copies text from another buffer to the
5321 current buffer, just the reverse---that is why the
5322 @code{append-to-buffer} definition starts out with a @code{let} that
5323 binds the local symbol @code{oldbuf} to the value returned by
5324 @code{current-buffer}.
5327 The @code{insert-buffer-substring} expression looks like this:
5330 (insert-buffer-substring oldbuf start end)
5334 The @code{insert-buffer-substring} function copies a string
5335 @emph{from} the buffer specified as its first argument and inserts the
5336 string into the present buffer. In this case, the argument to
5337 @code{insert-buffer-substring} is the value of the variable created
5338 and bound by the @code{let}, namely the value of @code{oldbuf}, which
5339 was the current buffer when you gave the @code{append-to-buffer}
5342 After @code{insert-buffer-substring} has done its work,
5343 @code{save-excursion} will restore the action to the original buffer
5344 and @code{append-to-buffer} will have done its job.
5347 Written in skeletal form, the workings of the body look like this:
5351 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5352 (save-excursion ; @r{Keep track of buffer.}
5354 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})
5356 @var{change-back-to-original-buffer-when-finished}
5357 @var{let-the-local-meaning-of-}@code{oldbuf}@var{-disappear-when-finished}
5361 In summary, @code{append-to-buffer} works as follows: it saves the
5362 value of the current buffer in the variable called @code{oldbuf}. It
5363 gets the new buffer (creating one if need be) and switches Emacs's
5364 attention to it. Using the value of @code{oldbuf}, it inserts the
5365 region of text from the old buffer into the new buffer; and then using
5366 @code{save-excursion}, it brings you back to your original buffer.
5368 In looking at @code{append-to-buffer}, you have explored a fairly
5369 complex function. It shows how to use @code{let} and
5370 @code{save-excursion}, and how to change to and come back from another
5371 buffer. Many function definitions use @code{let},
5372 @code{save-excursion}, and @code{set-buffer} this way.
5374 @node Buffer Related Review
5377 Here is a brief summary of the various functions discussed in this chapter.
5380 @item describe-function
5381 @itemx describe-variable
5382 Print the documentation for a function or variable.
5383 Conventionally bound to @kbd{C-h f} and @kbd{C-h v}.
5386 Find the file containing the source for a function or variable and
5387 switch buffers to it, positioning point at the beginning of the item.
5388 Conventionally bound to @kbd{M-.} (that's a period following the
5391 @item save-excursion
5392 Save the location of point and mark and restore their values after the
5393 arguments to @code{save-excursion} have been evaluated. Also, remember
5394 the current buffer and return to it.
5397 Set mark at a location and record the value of the previous mark on the
5398 mark ring. The mark is a location in the buffer that will keep its
5399 relative position even if text is added to or removed from the buffer.
5402 Set point to the location specified by the value of the argument, which
5403 can be a number, a marker, or an expression that returns the number of
5404 a position, such as @code{(point-min)}.
5406 @item insert-buffer-substring
5407 Copy a region of text from a buffer that is passed to the function as
5408 an argument and insert the region into the current buffer.
5410 @item mark-whole-buffer
5411 Mark the whole buffer as a region. Normally bound to @kbd{C-x h}.
5414 Switch the attention of Emacs to another buffer, but do not change the
5415 window being displayed. Used when the program rather than a human is
5416 to work on a different buffer.
5418 @item get-buffer-create
5420 Find a named buffer or create one if a buffer of that name does not
5421 exist. The @code{get-buffer} function returns @code{nil} if the named
5422 buffer does not exist.
5426 @node Buffer Exercises
5431 Write your own @code{simplified-end-of-buffer} function definition;
5432 then test it to see whether it works.
5435 Use @code{if} and @code{get-buffer} to write a function that prints a
5436 message telling you whether a buffer exists.
5439 Using @code{find-tag}, find the source for the @code{copy-to-buffer}
5444 @chapter A Few More Complex Functions
5446 In this chapter, we build on what we have learned in previous chapters
5447 by looking at more complex functions. The @code{copy-to-buffer}
5448 function illustrates use of two @code{save-excursion} expressions in
5449 one definition, while the @code{insert-buffer} function illustrates
5450 use of an asterisk in an @code{interactive} expression, use of
5451 @code{or}, and the important distinction between a name and the object
5452 to which the name refers.
5455 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
5456 * insert-buffer:: Read-only, and with @code{or}.
5457 * beginning-of-buffer:: Shows @code{goto-char},
5458 @code{point-min}, and @code{push-mark}.
5459 * Second Buffer Related Review::
5460 * optional Exercise::
5463 @node copy-to-buffer
5464 @section The Definition of @code{copy-to-buffer}
5465 @findex copy-to-buffer
5467 After understanding how @code{append-to-buffer} works, it is easy to
5468 understand @code{copy-to-buffer}. This function copies text into a
5469 buffer, but instead of adding to the second buffer, it replaces all the
5470 previous text in the second buffer.
5473 The body of @code{copy-to-buffer} looks like this,
5478 (interactive "BCopy to buffer: \nr")
5479 (let ((oldbuf (current-buffer)))
5480 (with-current-buffer (get-buffer-create buffer)
5481 (barf-if-buffer-read-only)
5484 (insert-buffer-substring oldbuf start end)))))
5488 The @code{copy-to-buffer} function has a simpler @code{interactive}
5489 expression than @code{append-to-buffer}.
5492 The definition then says
5495 (with-current-buffer (get-buffer-create buffer) @dots{}
5498 First, look at the earliest inner expression; that is evaluated first.
5499 That expression starts with @code{get-buffer-create buffer}. The
5500 function tells the computer to use the buffer with the name specified
5501 as the one to which you are copying, or if such a buffer does not
5502 exist, to create it. Then, the @code{with-current-buffer} function
5503 evaluates its body with that buffer temporarily current.
5505 (This demonstrates another way to shift the computer's attention but
5506 not the user's. The @code{append-to-buffer} function showed how to do
5507 the same with @code{save-excursion} and @code{set-buffer}.
5508 @code{with-current-buffer} is a newer, and arguably easier,
5511 The @code{barf-if-buffer-read-only} function sends you an error
5512 message saying the buffer is read-only if you cannot modify it.
5514 The next line has the @code{erase-buffer} function as its sole
5515 contents. That function erases the buffer.
5517 Finally, the last two lines contain the @code{save-excursion}
5518 expression with @code{insert-buffer-substring} as its body.
5519 The @code{insert-buffer-substring} expression copies the text from
5520 the buffer you are in (and you have not seen the computer shift its
5521 attention, so you don't know that that buffer is now called
5524 Incidentally, this is what is meant by `replacement'. To replace text,
5525 Emacs erases the previous text and then inserts new text.
5528 In outline, the body of @code{copy-to-buffer} looks like this:
5532 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5533 (@var{with-the-buffer-you-are-copying-to}
5534 (@var{but-do-not-erase-or-copy-to-a-read-only-buffer})
5537 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})))
5542 @section The Definition of @code{insert-buffer}
5543 @findex insert-buffer
5545 @code{insert-buffer} is yet another buffer-related function. This
5546 command copies another buffer @emph{into} the current buffer. It is the
5547 reverse of @code{append-to-buffer} or @code{copy-to-buffer}, since they
5548 copy a region of text @emph{from} the current buffer to another buffer.
5550 Here is a discussion based on the original code. The code was
5551 simplified in 2003 and is harder to understand.
5553 (@xref{New insert-buffer, , New Body for @code{insert-buffer}}, to see
5554 a discussion of the new body.)
5556 In addition, this code illustrates the use of @code{interactive} with a
5557 buffer that might be @dfn{read-only} and the important distinction
5558 between the name of an object and the object actually referred to.
5561 * insert-buffer code::
5562 * insert-buffer interactive:: When you can read, but not write.
5563 * insert-buffer body:: The body has an @code{or} and a @code{let}.
5564 * if & or:: Using an @code{if} instead of an @code{or}.
5565 * Insert or:: How the @code{or} expression works.
5566 * Insert let:: Two @code{save-excursion} expressions.
5567 * New insert-buffer::
5571 @node insert-buffer code
5572 @unnumberedsubsec The Code for @code{insert-buffer}
5576 Here is the earlier code:
5580 (defun insert-buffer (buffer)
5581 "Insert after point the contents of BUFFER.
5582 Puts mark after the inserted text.
5583 BUFFER may be a buffer or a buffer name."
5584 (interactive "*bInsert buffer:@: ")
5587 (or (bufferp buffer)
5588 (setq buffer (get-buffer buffer)))
5589 (let (start end newmark)
5593 (setq start (point-min) end (point-max)))
5596 (insert-buffer-substring buffer start end)
5597 (setq newmark (point)))
5598 (push-mark newmark)))
5603 As with other function definitions, you can use a template to see an
5604 outline of the function:
5608 (defun insert-buffer (buffer)
5609 "@var{documentation}@dots{}"
5610 (interactive "*bInsert buffer:@: ")
5615 @node insert-buffer interactive
5616 @subsection The Interactive Expression in @code{insert-buffer}
5617 @findex interactive, @r{example use of}
5619 In @code{insert-buffer}, the argument to the @code{interactive}
5620 declaration has two parts, an asterisk, @samp{*}, and @samp{bInsert
5624 * Read-only buffer:: When a buffer cannot be modified.
5625 * b for interactive:: An existing buffer or else its name.
5628 @node Read-only buffer
5629 @unnumberedsubsubsec A Read-only Buffer
5630 @cindex Read-only buffer
5631 @cindex Asterisk for read-only buffer
5632 @findex * @r{for read-only buffer}
5634 The asterisk is for the situation when the current buffer is a
5635 read-only buffer---a buffer that cannot be modified. If
5636 @code{insert-buffer} is called when the current buffer is read-only, a
5637 message to this effect is printed in the echo area and the terminal
5638 may beep or blink at you; you will not be permitted to insert anything
5639 into current buffer. The asterisk does not need to be followed by a
5640 newline to separate it from the next argument.
5642 @node b for interactive
5643 @unnumberedsubsubsec @samp{b} in an Interactive Expression
5645 The next argument in the interactive expression starts with a lower
5646 case @samp{b}. (This is different from the code for
5647 @code{append-to-buffer}, which uses an upper-case @samp{B}.
5648 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
5649 The lower-case @samp{b} tells the Lisp interpreter that the argument
5650 for @code{insert-buffer} should be an existing buffer or else its
5651 name. (The upper-case @samp{B} option provides for the possibility
5652 that the buffer does not exist.) Emacs will prompt you for the name
5653 of the buffer, offering you a default buffer, with name completion
5654 enabled. If the buffer does not exist, you receive a message that
5655 says ``No match''; your terminal may beep at you as well.
5657 The new and simplified code generates a list for @code{interactive}.
5658 It uses the @code{barf-if-buffer-read-only} and @code{read-buffer}
5659 functions with which we are already familiar and the @code{progn}
5660 special form with which we are not. (It will be described later.)
5662 @node insert-buffer body
5663 @subsection The Body of the @code{insert-buffer} Function
5665 The body of the @code{insert-buffer} function has two major parts: an
5666 @code{or} expression and a @code{let} expression. The purpose of the
5667 @code{or} expression is to ensure that the argument @code{buffer} is
5668 bound to a buffer and not just the name of a buffer. The body of the
5669 @code{let} expression contains the code which copies the other buffer
5670 into the current buffer.
5673 In outline, the two expressions fit into the @code{insert-buffer}
5678 (defun insert-buffer (buffer)
5679 "@var{documentation}@dots{}"
5680 (interactive "*bInsert buffer:@: ")
5685 (let (@var{varlist})
5686 @var{body-of-}@code{let}@dots{} )
5690 To understand how the @code{or} expression ensures that the argument
5691 @code{buffer} is bound to a buffer and not to the name of a buffer, it
5692 is first necessary to understand the @code{or} function.
5694 Before doing this, let me rewrite this part of the function using
5695 @code{if} so that you can see what is done in a manner that will be familiar.
5698 @subsection @code{insert-buffer} With an @code{if} Instead of an @code{or}
5700 The job to be done is to make sure the value of @code{buffer} is a
5701 buffer itself and not the name of a buffer. If the value is the name,
5702 then the buffer itself must be got.
5704 You can imagine yourself at a conference where an usher is wandering
5705 around holding a list with your name on it and looking for you: the
5706 usher is ``bound'' to your name, not to you; but when the usher finds
5707 you and takes your arm, the usher becomes ``bound'' to you.
5710 In Lisp, you might describe this situation like this:
5714 (if (not (holding-on-to-guest))
5715 (find-and-take-arm-of-guest))
5719 We want to do the same thing with a buffer---if we do not have the
5720 buffer itself, we want to get it.
5723 Using a predicate called @code{bufferp} that tells us whether we have a
5724 buffer (rather than its name), we can write the code like this:
5728 (if (not (bufferp buffer)) ; @r{if-part}
5729 (setq buffer (get-buffer buffer))) ; @r{then-part}
5734 Here, the true-or-false-test of the @code{if} expression is
5735 @w{@code{(not (bufferp buffer))}}; and the then-part is the expression
5736 @w{@code{(setq buffer (get-buffer buffer))}}.
5738 In the test, the function @code{bufferp} returns true if its argument is
5739 a buffer---but false if its argument is the name of the buffer. (The
5740 last character of the function name @code{bufferp} is the character
5741 @samp{p}; as we saw earlier, such use of @samp{p} is a convention that
5742 indicates that the function is a predicate, which is a term that means
5743 that the function will determine whether some property is true or false.
5744 @xref{Wrong Type of Argument, , Using the Wrong Type Object as an
5748 The function @code{not} precedes the expression @code{(bufferp buffer)},
5749 so the true-or-false-test looks like this:
5752 (not (bufferp buffer))
5756 @code{not} is a function that returns true if its argument is false
5757 and false if its argument is true. So if @code{(bufferp buffer)}
5758 returns true, the @code{not} expression returns false and vice-verse:
5759 what is ``not true'' is false and what is ``not false'' is true.
5761 Using this test, the @code{if} expression works as follows: when the
5762 value of the variable @code{buffer} is actually a buffer rather than
5763 its name, the true-or-false-test returns false and the @code{if}
5764 expression does not evaluate the then-part. This is fine, since we do
5765 not need to do anything to the variable @code{buffer} if it really is
5768 On the other hand, when the value of @code{buffer} is not a buffer
5769 itself, but the name of a buffer, the true-or-false-test returns true
5770 and the then-part of the expression is evaluated. In this case, the
5771 then-part is @code{(setq buffer (get-buffer buffer))}. This
5772 expression uses the @code{get-buffer} function to return an actual
5773 buffer itself, given its name. The @code{setq} then sets the variable
5774 @code{buffer} to the value of the buffer itself, replacing its previous
5775 value (which was the name of the buffer).
5778 @subsection The @code{or} in the Body
5780 The purpose of the @code{or} expression in the @code{insert-buffer}
5781 function is to ensure that the argument @code{buffer} is bound to a
5782 buffer and not just to the name of a buffer. The previous section shows
5783 how the job could have been done using an @code{if} expression.
5784 However, the @code{insert-buffer} function actually uses @code{or}.
5785 To understand this, it is necessary to understand how @code{or} works.
5788 An @code{or} function can have any number of arguments. It evaluates
5789 each argument in turn and returns the value of the first of its
5790 arguments that is not @code{nil}. Also, and this is a crucial feature
5791 of @code{or}, it does not evaluate any subsequent arguments after
5792 returning the first non-@code{nil} value.
5795 The @code{or} expression looks like this:
5799 (or (bufferp buffer)
5800 (setq buffer (get-buffer buffer)))
5805 The first argument to @code{or} is the expression @code{(bufferp buffer)}.
5806 This expression returns true (a non-@code{nil} value) if the buffer is
5807 actually a buffer, and not just the name of a buffer. In the @code{or}
5808 expression, if this is the case, the @code{or} expression returns this
5809 true value and does not evaluate the next expression---and this is fine
5810 with us, since we do not want to do anything to the value of
5811 @code{buffer} if it really is a buffer.
5813 On the other hand, if the value of @code{(bufferp buffer)} is @code{nil},
5814 which it will be if the value of @code{buffer} is the name of a buffer,
5815 the Lisp interpreter evaluates the next element of the @code{or}
5816 expression. This is the expression @code{(setq buffer (get-buffer
5817 buffer))}. This expression returns a non-@code{nil} value, which
5818 is the value to which it sets the variable @code{buffer}---and this
5819 value is a buffer itself, not the name of a buffer.
5821 The result of all this is that the symbol @code{buffer} is always
5822 bound to a buffer itself rather than to the name of a buffer. All
5823 this is necessary because the @code{set-buffer} function in a
5824 following line only works with a buffer itself, not with the name to a
5828 Incidentally, using @code{or}, the situation with the usher would be
5832 (or (holding-on-to-guest) (find-and-take-arm-of-guest))
5836 @subsection The @code{let} Expression in @code{insert-buffer}
5838 After ensuring that the variable @code{buffer} refers to a buffer itself
5839 and not just to the name of a buffer, the @code{insert-buffer function}
5840 continues with a @code{let} expression. This specifies three local
5841 variables, @code{start}, @code{end}, and @code{newmark} and binds them
5842 to the initial value @code{nil}. These variables are used inside the
5843 remainder of the @code{let} and temporarily hide any other occurrence of
5844 variables of the same name in Emacs until the end of the @code{let}.
5847 The body of the @code{let} contains two @code{save-excursion}
5848 expressions. First, we will look at the inner @code{save-excursion}
5849 expression in detail. The expression looks like this:
5855 (setq start (point-min) end (point-max)))
5860 The expression @code{(set-buffer buffer)} changes Emacs's attention
5861 from the current buffer to the one from which the text will copied.
5862 In that buffer, the variables @code{start} and @code{end} are set to
5863 the beginning and end of the buffer, using the commands
5864 @code{point-min} and @code{point-max}. Note that we have here an
5865 illustration of how @code{setq} is able to set two variables in the
5866 same expression. The first argument of @code{setq} is set to the
5867 value of its second, and its third argument is set to the value of its
5870 After the body of the inner @code{save-excursion} is evaluated, the
5871 @code{save-excursion} restores the original buffer, but @code{start} and
5872 @code{end} remain set to the values of the beginning and end of the
5873 buffer from which the text will be copied.
5876 The outer @code{save-excursion} expression looks like this:
5881 (@var{inner-}@code{save-excursion}@var{-expression}
5882 (@var{go-to-new-buffer-and-set-}@code{start}@var{-and-}@code{end})
5883 (insert-buffer-substring buffer start end)
5884 (setq newmark (point)))
5889 The @code{insert-buffer-substring} function copies the text
5890 @emph{into} the current buffer @emph{from} the region indicated by
5891 @code{start} and @code{end} in @code{buffer}. Since the whole of the
5892 second buffer lies between @code{start} and @code{end}, the whole of
5893 the second buffer is copied into the buffer you are editing. Next,
5894 the value of point, which will be at the end of the inserted text, is
5895 recorded in the variable @code{newmark}.
5897 After the body of the outer @code{save-excursion} is evaluated, point
5898 and mark are relocated to their original places.
5900 However, it is convenient to locate a mark at the end of the newly
5901 inserted text and locate point at its beginning. The @code{newmark}
5902 variable records the end of the inserted text. In the last line of
5903 the @code{let} expression, the @code{(push-mark newmark)} expression
5904 function sets a mark to this location. (The previous location of the
5905 mark is still accessible; it is recorded on the mark ring and you can
5906 go back to it with @kbd{C-u C-@key{SPC}}.) Meanwhile, point is
5907 located at the beginning of the inserted text, which is where it was
5908 before you called the insert function, the position of which was saved
5909 by the first @code{save-excursion}.
5912 The whole @code{let} expression looks like this:
5916 (let (start end newmark)
5920 (setq start (point-min) end (point-max)))
5921 (insert-buffer-substring buffer start end)
5922 (setq newmark (point)))
5923 (push-mark newmark))
5927 Like the @code{append-to-buffer} function, the @code{insert-buffer}
5928 function uses @code{let}, @code{save-excursion}, and
5929 @code{set-buffer}. In addition, the function illustrates one way to
5930 use @code{or}. All these functions are building blocks that we will
5931 find and use again and again.
5933 @node New insert-buffer
5934 @subsection New Body for @code{insert-buffer}
5935 @findex insert-buffer, new version body
5936 @findex new version body for insert-buffer
5938 The body in the GNU Emacs 22 version is more confusing than the original.
5941 It consists of two expressions,
5947 (insert-buffer-substring (get-buffer buffer))
5955 except, and this is what confuses novices, very important work is done
5956 inside the @code{push-mark} expression.
5958 The @code{get-buffer} function returns a buffer with the name
5959 provided. You will note that the function is @emph{not} called
5960 @code{get-buffer-create}; it does not create a buffer if one does not
5961 already exist. The buffer returned by @code{get-buffer}, an existing
5962 buffer, is passed to @code{insert-buffer-substring}, which inserts the
5963 whole of the buffer (since you did not specify anything else).
5965 The location into which the buffer is inserted is recorded by
5966 @code{push-mark}. Then the function returns @code{nil}, the value of
5967 its last command. Put another way, the @code{insert-buffer} function
5968 exists only to produce a side effect, inserting another buffer, not to
5971 @node beginning-of-buffer
5972 @section Complete Definition of @code{beginning-of-buffer}
5973 @findex beginning-of-buffer
5975 The basic structure of the @code{beginning-of-buffer} function has
5976 already been discussed. (@xref{simplified-beginning-of-buffer, , A
5977 Simplified @code{beginning-of-buffer} Definition}.)
5978 This section describes the complex part of the definition.
5980 As previously described, when invoked without an argument,
5981 @code{beginning-of-buffer} moves the cursor to the beginning of the
5982 buffer (in truth, the beginning of the accessible portion of the
5983 buffer), leaving the mark at the previous position. However, when the
5984 command is invoked with a number between one and ten, the function
5985 considers that number to be a fraction of the length of the buffer,
5986 measured in tenths, and Emacs moves the cursor that fraction of the
5987 way from the beginning of the buffer. Thus, you can either call this
5988 function with the key command @kbd{M-<}, which will move the cursor to
5989 the beginning of the buffer, or with a key command such as @kbd{C-u 7
5990 M-<} which will move the cursor to a point 70% of the way through the
5991 buffer. If a number bigger than ten is used for the argument, it
5992 moves to the end of the buffer.
5994 The @code{beginning-of-buffer} function can be called with or without an
5995 argument. The use of the argument is optional.
5998 * Optional Arguments::
5999 * beginning-of-buffer opt arg:: Example with optional argument.
6000 * beginning-of-buffer complete::
6003 @node Optional Arguments
6004 @subsection Optional Arguments
6006 Unless told otherwise, Lisp expects that a function with an argument in
6007 its function definition will be called with a value for that argument.
6008 If that does not happen, you get an error and a message that says
6009 @samp{Wrong number of arguments}.
6011 @cindex Optional arguments
6014 However, optional arguments are a feature of Lisp: a particular
6015 @dfn{keyword} is used to tell the Lisp interpreter that an argument is
6016 optional. The keyword is @code{&optional}. (The @samp{&} in front of
6017 @samp{optional} is part of the keyword.) In a function definition, if
6018 an argument follows the keyword @code{&optional}, no value need be
6019 passed to that argument when the function is called.
6022 The first line of the function definition of @code{beginning-of-buffer}
6023 therefore looks like this:
6026 (defun beginning-of-buffer (&optional arg)
6030 In outline, the whole function looks like this:
6034 (defun beginning-of-buffer (&optional arg)
6035 "@var{documentation}@dots{}"
6037 (or (@var{is-the-argument-a-cons-cell} arg)
6038 (and @var{are-both-transient-mark-mode-and-mark-active-true})
6040 (let (@var{determine-size-and-set-it})
6042 (@var{if-there-is-an-argument}
6043 @var{figure-out-where-to-go}
6050 The function is similar to the @code{simplified-beginning-of-buffer}
6051 function except that the @code{interactive} expression has @code{"P"}
6052 as an argument and the @code{goto-char} function is followed by an
6053 if-then-else expression that figures out where to put the cursor if
6054 there is an argument that is not a cons cell.
6056 (Since I do not explain a cons cell for many more chapters, please
6057 consider ignoring the function @code{consp}. @xref{List
6058 Implementation, , How Lists are Implemented}, and @ref{Cons Cell Type,
6059 , Cons Cell and List Types, elisp, The GNU Emacs Lisp Reference
6062 The @code{"P"} in the @code{interactive} expression tells Emacs to
6063 pass a prefix argument, if there is one, to the function in raw form.
6064 A prefix argument is made by typing the @key{META} key followed by a
6065 number, or by typing @kbd{C-u} and then a number. (If you don't type
6066 a number, @kbd{C-u} defaults to a cons cell with a 4. A lowercase
6067 @code{"p"} in the @code{interactive} expression causes the function to
6068 convert a prefix arg to a number.)
6070 The true-or-false-test of the @code{if} expression looks complex, but
6071 it is not: it checks whether @code{arg} has a value that is not
6072 @code{nil} and whether it is a cons cell. (That is what @code{consp}
6073 does; it checks whether its argument is a cons cell.) If @code{arg}
6074 has a value that is not @code{nil} (and is not a cons cell), which
6075 will be the case if @code{beginning-of-buffer} is called with a
6076 numeric argument, then this true-or-false-test will return true and
6077 the then-part of the @code{if} expression will be evaluated. On the
6078 other hand, if @code{beginning-of-buffer} is not called with an
6079 argument, the value of @code{arg} will be @code{nil} and the else-part
6080 of the @code{if} expression will be evaluated. The else-part is
6081 simply @code{point-min}, and when this is the outcome, the whole
6082 @code{goto-char} expression is @code{(goto-char (point-min))}, which
6083 is how we saw the @code{beginning-of-buffer} function in its
6086 @node beginning-of-buffer opt arg
6087 @subsection @code{beginning-of-buffer} with an Argument
6089 When @code{beginning-of-buffer} is called with an argument, an
6090 expression is evaluated which calculates what value to pass to
6091 @code{goto-char}. This expression is rather complicated at first sight.
6092 It includes an inner @code{if} expression and much arithmetic. It looks
6097 (if (> (buffer-size) 10000)
6098 ;; @r{Avoid overflow for large buffer sizes!}
6099 (* (prefix-numeric-value arg)
6104 size (prefix-numeric-value arg))) 10)))
6109 * Disentangle beginning-of-buffer::
6110 * Large buffer case::
6111 * Small buffer case::
6115 @node Disentangle beginning-of-buffer
6116 @unnumberedsubsubsec Disentangle @code{beginning-of-buffer}
6119 Like other complex-looking expressions, the conditional expression
6120 within @code{beginning-of-buffer} can be disentangled by looking at it
6121 as parts of a template, in this case, the template for an if-then-else
6122 expression. In skeletal form, the expression looks like this:
6126 (if (@var{buffer-is-large}
6127 @var{divide-buffer-size-by-10-and-multiply-by-arg}
6128 @var{else-use-alternate-calculation}
6132 The true-or-false-test of this inner @code{if} expression checks the
6133 size of the buffer. The reason for this is that the old version 18
6134 Emacs used numbers that are no bigger than eight million or so and in
6135 the computation that followed, the programmer feared that Emacs might
6136 try to use over-large numbers if the buffer were large. The term
6137 `overflow', mentioned in the comment, means numbers that are over
6138 large. More recent versions of Emacs use larger numbers, but this
6139 code has not been touched, if only because people now look at buffers
6140 that are far, far larger than ever before.
6142 There are two cases: if the buffer is large and if it is not.
6144 @node Large buffer case
6145 @unnumberedsubsubsec What happens in a large buffer
6147 In @code{beginning-of-buffer}, the inner @code{if} expression tests
6148 whether the size of the buffer is greater than 10,000 characters. To do
6149 this, it uses the @code{>} function and the computation of @code{size}
6150 that comes from the let expression.
6152 In the old days, the function @code{buffer-size} was used. Not only
6153 was that function called several times, it gave the size of the whole
6154 buffer, not the accessible part. The computation makes much more
6155 sense when it handles just the accessible part. (@xref{Narrowing &
6156 Widening, , Narrowing and Widening}, for more information on focusing
6157 attention to an `accessible' part.)
6160 The line looks like this:
6168 When the buffer is large, the then-part of the @code{if} expression is
6169 evaluated. It reads like this (after formatting for easy reading):
6174 (prefix-numeric-value arg)
6180 This expression is a multiplication, with two arguments to the function
6183 The first argument is @code{(prefix-numeric-value arg)}. When
6184 @code{"P"} is used as the argument for @code{interactive}, the value
6185 passed to the function as its argument is passed a ``raw prefix
6186 argument'', and not a number. (It is a number in a list.) To perform
6187 the arithmetic, a conversion is necessary, and
6188 @code{prefix-numeric-value} does the job.
6190 @findex / @r{(division)}
6192 The second argument is @code{(/ size 10)}. This expression divides
6193 the numeric value by ten---the numeric value of the size of the
6194 accessible portion of the buffer. This produces a number that tells
6195 how many characters make up one tenth of the buffer size. (In Lisp,
6196 @code{/} is used for division, just as @code{*} is used for
6200 In the multiplication expression as a whole, this amount is multiplied
6201 by the value of the prefix argument---the multiplication looks like this:
6205 (* @var{numeric-value-of-prefix-arg}
6206 @var{number-of-characters-in-one-tenth-of-the-accessible-buffer})
6211 If, for example, the prefix argument is @samp{7}, the one-tenth value
6212 will be multiplied by 7 to give a position 70% of the way through.
6215 The result of all this is that if the accessible portion of the buffer
6216 is large, the @code{goto-char} expression reads like this:
6220 (goto-char (* (prefix-numeric-value arg)
6225 This puts the cursor where we want it.
6227 @node Small buffer case
6228 @unnumberedsubsubsec What happens in a small buffer
6230 If the buffer contains fewer than 10,000 characters, a slightly
6231 different computation is performed. You might think this is not
6232 necessary, since the first computation could do the job. However, in
6233 a small buffer, the first method may not put the cursor on exactly the
6234 desired line; the second method does a better job.
6237 The code looks like this:
6239 @c Keep this on one line.
6241 (/ (+ 10 (* size (prefix-numeric-value arg))) 10))
6246 This is code in which you figure out what happens by discovering how the
6247 functions are embedded in parentheses. It is easier to read if you
6248 reformat it with each expression indented more deeply than its
6249 enclosing expression:
6257 (prefix-numeric-value arg)))
6264 Looking at parentheses, we see that the innermost operation is
6265 @code{(prefix-numeric-value arg)}, which converts the raw argument to
6266 a number. In the following expression, this number is multiplied by
6267 the size of the accessible portion of the buffer:
6270 (* size (prefix-numeric-value arg))
6274 This multiplication creates a number that may be larger than the size of
6275 the buffer---seven times larger if the argument is 7, for example. Ten
6276 is then added to this number and finally the large number is divided by
6277 ten to provide a value that is one character larger than the percentage
6278 position in the buffer.
6280 The number that results from all this is passed to @code{goto-char} and
6281 the cursor is moved to that point.
6284 @node beginning-of-buffer complete
6285 @subsection The Complete @code{beginning-of-buffer}
6288 Here is the complete text of the @code{beginning-of-buffer} function:
6294 (defun beginning-of-buffer (&optional arg)
6295 "Move point to the beginning of the buffer;
6296 leave mark at previous position.
6297 With \\[universal-argument] prefix,
6298 do not set mark at previous position.
6300 put point N/10 of the way from the beginning.
6302 If the buffer is narrowed,
6303 this command uses the beginning and size
6304 of the accessible part of the buffer.
6308 Don't use this command in Lisp programs!
6309 \(goto-char (point-min)) is faster
6310 and avoids clobbering the mark."
6313 (and transient-mark-mode mark-active)
6317 (let ((size (- (point-max) (point-min))))
6318 (goto-char (if (and arg (not (consp arg)))
6321 ;; Avoid overflow for large buffer sizes!
6322 (* (prefix-numeric-value arg)
6324 (/ (+ 10 (* size (prefix-numeric-value arg)))
6327 (if (and arg (not (consp arg))) (forward-line 1)))
6332 From before GNU Emacs 22
6335 (defun beginning-of-buffer (&optional arg)
6336 "Move point to the beginning of the buffer;
6337 leave mark at previous position.
6338 With arg N, put point N/10 of the way
6339 from the true beginning.
6342 Don't use this in Lisp programs!
6343 \(goto-char (point-min)) is faster
6344 and does not set the mark."
6351 (if (> (buffer-size) 10000)
6352 ;; @r{Avoid overflow for large buffer sizes!}
6353 (* (prefix-numeric-value arg)
6354 (/ (buffer-size) 10))
6357 (/ (+ 10 (* (buffer-size)
6358 (prefix-numeric-value arg)))
6361 (if arg (forward-line 1)))
6367 Except for two small points, the previous discussion shows how this
6368 function works. The first point deals with a detail in the
6369 documentation string, and the second point concerns the last line of
6373 In the documentation string, there is reference to an expression:
6376 \\[universal-argument]
6380 A @samp{\\} is used before the first square bracket of this
6381 expression. This @samp{\\} tells the Lisp interpreter to substitute
6382 whatever key is currently bound to the @samp{[@dots{}]}. In the case
6383 of @code{universal-argument}, that is usually @kbd{C-u}, but it might
6384 be different. (@xref{Documentation Tips, , Tips for Documentation
6385 Strings, elisp, The GNU Emacs Lisp Reference Manual}, for more
6389 Finally, the last line of the @code{beginning-of-buffer} command says
6390 to move point to the beginning of the next line if the command is
6391 invoked with an argument:
6394 (if (and arg (not (consp arg))) (forward-line 1))
6398 This puts the cursor at the beginning of the first line after the
6399 appropriate tenths position in the buffer. This is a flourish that
6400 means that the cursor is always located @emph{at least} the requested
6401 tenths of the way through the buffer, which is a nicety that is,
6402 perhaps, not necessary, but which, if it did not occur, would be sure
6403 to draw complaints. (The @code{(not (consp arg))} portion is so that
6404 if you specify the command with a @kbd{C-u}, but without a number,
6405 that is to say, if the `raw prefix argument' is simply a cons cell,
6406 the command does not put you at the beginning of the second line.)
6408 @node Second Buffer Related Review
6411 Here is a brief summary of some of the topics covered in this chapter.
6415 Evaluate each argument in sequence, and return the value of the first
6416 argument that is not @code{nil}; if none return a value that is not
6417 @code{nil}, return @code{nil}. In brief, return the first true value
6418 of the arguments; return a true value if one @emph{or} any of the
6422 Evaluate each argument in sequence, and if any are @code{nil}, return
6423 @code{nil}; if none are @code{nil}, return the value of the last
6424 argument. In brief, return a true value only if all the arguments are
6425 true; return a true value if one @emph{and} each of the others is
6429 A keyword used to indicate that an argument to a function definition
6430 is optional; this means that the function can be evaluated without the
6431 argument, if desired.
6433 @item prefix-numeric-value
6434 Convert the `raw prefix argument' produced by @code{(interactive
6435 "P")} to a numeric value.
6438 Move point forward to the beginning of the next line, or if the argument
6439 is greater than one, forward that many lines. If it can't move as far
6440 forward as it is supposed to, @code{forward-line} goes forward as far as
6441 it can and then returns a count of the number of additional lines it was
6442 supposed to move but couldn't.
6445 Delete the entire contents of the current buffer.
6448 Return @code{t} if its argument is a buffer; otherwise return @code{nil}.
6451 @node optional Exercise
6452 @section @code{optional} Argument Exercise
6454 Write an interactive function with an optional argument that tests
6455 whether its argument, a number, is greater than or equal to, or else,
6456 less than the value of @code{fill-column}, and tells you which, in a
6457 message. However, if you do not pass an argument to the function, use
6458 56 as a default value.
6460 @node Narrowing & Widening
6461 @chapter Narrowing and Widening
6462 @cindex Focusing attention (narrowing)
6466 Narrowing is a feature of Emacs that makes it possible for you to focus
6467 on a specific part of a buffer, and work without accidentally changing
6468 other parts. Narrowing is normally disabled since it can confuse
6472 * Narrowing advantages:: The advantages of narrowing
6473 * save-restriction:: The @code{save-restriction} special form.
6474 * what-line:: The number of the line that point is on.
6479 @node Narrowing advantages
6480 @unnumberedsec The Advantages of Narrowing
6483 With narrowing, the rest of a buffer is made invisible, as if it weren't
6484 there. This is an advantage if, for example, you want to replace a word
6485 in one part of a buffer but not in another: you narrow to the part you want
6486 and the replacement is carried out only in that section, not in the rest
6487 of the buffer. Searches will only work within a narrowed region, not
6488 outside of one, so if you are fixing a part of a document, you can keep
6489 yourself from accidentally finding parts you do not need to fix by
6490 narrowing just to the region you want.
6491 (The key binding for @code{narrow-to-region} is @kbd{C-x n n}.)
6493 However, narrowing does make the rest of the buffer invisible, which
6494 can scare people who inadvertently invoke narrowing and think they
6495 have deleted a part of their file. Moreover, the @code{undo} command
6496 (which is usually bound to @kbd{C-x u}) does not turn off narrowing
6497 (nor should it), so people can become quite desperate if they do not
6498 know that they can return the rest of a buffer to visibility with the
6499 @code{widen} command.
6500 (The key binding for @code{widen} is @kbd{C-x n w}.)
6502 Narrowing is just as useful to the Lisp interpreter as to a human.
6503 Often, an Emacs Lisp function is designed to work on just part of a
6504 buffer; or conversely, an Emacs Lisp function needs to work on all of a
6505 buffer that has been narrowed. The @code{what-line} function, for
6506 example, removes the narrowing from a buffer, if it has any narrowing
6507 and when it has finished its job, restores the narrowing to what it was.
6508 On the other hand, the @code{count-lines} function
6509 uses narrowing to restrict itself to just that portion
6510 of the buffer in which it is interested and then restores the previous
6513 @node save-restriction
6514 @section The @code{save-restriction} Special Form
6515 @findex save-restriction
6517 In Emacs Lisp, you can use the @code{save-restriction} special form to
6518 keep track of whatever narrowing is in effect, if any. When the Lisp
6519 interpreter meets with @code{save-restriction}, it executes the code
6520 in the body of the @code{save-restriction} expression, and then undoes
6521 any changes to narrowing that the code caused. If, for example, the
6522 buffer is narrowed and the code that follows @code{save-restriction}
6523 gets rid of the narrowing, @code{save-restriction} returns the buffer
6524 to its narrowed region afterwards. In the @code{what-line} command,
6525 any narrowing the buffer may have is undone by the @code{widen}
6526 command that immediately follows the @code{save-restriction} command.
6527 Any original narrowing is restored just before the completion of the
6531 The template for a @code{save-restriction} expression is simple:
6541 The body of the @code{save-restriction} is one or more expressions that
6542 will be evaluated in sequence by the Lisp interpreter.
6544 Finally, a point to note: when you use both @code{save-excursion} and
6545 @code{save-restriction}, one right after the other, you should use
6546 @code{save-excursion} outermost. If you write them in reverse order,
6547 you may fail to record narrowing in the buffer to which Emacs switches
6548 after calling @code{save-excursion}. Thus, when written together,
6549 @code{save-excursion} and @code{save-restriction} should be written
6560 In other circumstances, when not written together, the
6561 @code{save-excursion} and @code{save-restriction} special forms must
6562 be written in the order appropriate to the function.
6578 /usr/local/src/emacs/lisp/simple.el
6581 "Print the current buffer line number and narrowed line number of point."
6583 (let ((start (point-min))
6584 (n (line-number-at-pos)))
6586 (message "Line %d" n)
6590 (message "line %d (narrowed line %d)"
6591 (+ n (line-number-at-pos start) -1) n))))))
6593 (defun line-number-at-pos (&optional pos)
6594 "Return (narrowed) buffer line number at position POS.
6595 If POS is nil, use current buffer location.
6596 Counting starts at (point-min), so the value refers
6597 to the contents of the accessible portion of the buffer."
6598 (let ((opoint (or pos (point))) start)
6600 (goto-char (point-min))
6601 (setq start (point))
6604 (1+ (count-lines start (point))))))
6606 (defun count-lines (start end)
6607 "Return number of lines between START and END.
6608 This is usually the number of newlines between them,
6609 but can be one more if START is not equal to END
6610 and the greater of them is not at the start of a line."
6613 (narrow-to-region start end)
6614 (goto-char (point-min))
6615 (if (eq selective-display t)
6618 (while (re-search-forward "[\n\C-m]" nil t 40)
6619 (setq done (+ 40 done)))
6620 (while (re-search-forward "[\n\C-m]" nil t 1)
6621 (setq done (+ 1 done)))
6622 (goto-char (point-max))
6623 (if (and (/= start end)
6627 (- (buffer-size) (forward-line (buffer-size)))))))
6631 @section @code{what-line}
6633 @cindex Widening, example of
6635 The @code{what-line} command tells you the number of the line in which
6636 the cursor is located. The function illustrates the use of the
6637 @code{save-restriction} and @code{save-excursion} commands. Here is the
6638 original text of the function:
6643 "Print the current line number (in the buffer) of point."
6650 (1+ (count-lines 1 (point)))))))
6654 (In recent versions of GNU Emacs, the @code{what-line} function has
6655 been expanded to tell you your line number in a narrowed buffer as
6656 well as your line number in a widened buffer. The recent version is
6657 more complex than the version shown here. If you feel adventurous,
6658 you might want to look at it after figuring out how this version
6659 works. You will probably need to use @kbd{C-h f}
6660 (@code{describe-function}). The newer version uses a conditional to
6661 determine whether the buffer has been narrowed.
6663 (Also, it uses @code{line-number-at-pos}, which among other simple
6664 expressions, such as @code{(goto-char (point-min))}, moves point to
6665 the beginning of the current line with @code{(forward-line 0)} rather
6666 than @code{beginning-of-line}.)
6668 The @code{what-line} function as shown here has a documentation line
6669 and is interactive, as you would expect. The next two lines use the
6670 functions @code{save-restriction} and @code{widen}.
6672 The @code{save-restriction} special form notes whatever narrowing is in
6673 effect, if any, in the current buffer and restores that narrowing after
6674 the code in the body of the @code{save-restriction} has been evaluated.
6676 The @code{save-restriction} special form is followed by @code{widen}.
6677 This function undoes any narrowing the current buffer may have had
6678 when @code{what-line} was called. (The narrowing that was there is
6679 the narrowing that @code{save-restriction} remembers.) This widening
6680 makes it possible for the line counting commands to count from the
6681 beginning of the buffer. Otherwise, they would have been limited to
6682 counting within the accessible region. Any original narrowing is
6683 restored just before the completion of the function by the
6684 @code{save-restriction} special form.
6686 The call to @code{widen} is followed by @code{save-excursion}, which
6687 saves the location of the cursor (i.e., of point) and of the mark, and
6688 restores them after the code in the body of the @code{save-excursion}
6689 uses the @code{beginning-of-line} function to move point.
6691 (Note that the @code{(widen)} expression comes between the
6692 @code{save-restriction} and @code{save-excursion} special forms. When
6693 you write the two @code{save- @dots{}} expressions in sequence, write
6694 @code{save-excursion} outermost.)
6697 The last two lines of the @code{what-line} function are functions to
6698 count the number of lines in the buffer and then print the number in the
6704 (1+ (count-lines 1 (point)))))))
6708 The @code{message} function prints a one-line message at the bottom of
6709 the Emacs screen. The first argument is inside of quotation marks and
6710 is printed as a string of characters. However, it may contain a
6711 @samp{%d} expression to print a following argument. @samp{%d} prints
6712 the argument as a decimal, so the message will say something such as
6716 The number that is printed in place of the @samp{%d} is computed by the
6717 last line of the function:
6720 (1+ (count-lines 1 (point)))
6726 (defun count-lines (start end)
6727 "Return number of lines between START and END.
6728 This is usually the number of newlines between them,
6729 but can be one more if START is not equal to END
6730 and the greater of them is not at the start of a line."
6733 (narrow-to-region start end)
6734 (goto-char (point-min))
6735 (if (eq selective-display t)
6738 (while (re-search-forward "[\n\C-m]" nil t 40)
6739 (setq done (+ 40 done)))
6740 (while (re-search-forward "[\n\C-m]" nil t 1)
6741 (setq done (+ 1 done)))
6742 (goto-char (point-max))
6743 (if (and (/= start end)
6747 (- (buffer-size) (forward-line (buffer-size)))))))
6751 What this does is count the lines from the first position of the
6752 buffer, indicated by the @code{1}, up to @code{(point)}, and then add
6753 one to that number. (The @code{1+} function adds one to its
6754 argument.) We add one to it because line 2 has only one line before
6755 it, and @code{count-lines} counts only the lines @emph{before} the
6758 After @code{count-lines} has done its job, and the message has been
6759 printed in the echo area, the @code{save-excursion} restores point and
6760 mark to their original positions; and @code{save-restriction} restores
6761 the original narrowing, if any.
6763 @node narrow Exercise
6764 @section Exercise with Narrowing
6766 Write a function that will display the first 60 characters of the
6767 current buffer, even if you have narrowed the buffer to its latter
6768 half so that the first line is inaccessible. Restore point, mark, and
6769 narrowing. For this exercise, you need to use a whole potpourri of
6770 functions, including @code{save-restriction}, @code{widen},
6771 @code{goto-char}, @code{point-min}, @code{message}, and
6772 @code{buffer-substring}.
6774 @cindex Properties, mention of @code{buffer-substring-no-properties}
6775 (@code{buffer-substring} is a previously unmentioned function you will
6776 have to investigate yourself; or perhaps you will have to use
6777 @code{buffer-substring-no-properties} or
6778 @code{filter-buffer-substring} @dots{}, yet other functions. Text
6779 properties are a feature otherwise not discussed here. @xref{Text
6780 Properties, , Text Properties, elisp, The GNU Emacs Lisp Reference
6783 Additionally, do you really need @code{goto-char} or @code{point-min}?
6784 Or can you write the function without them?
6786 @node car cdr & cons
6787 @chapter @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
6788 @findex car, @r{introduced}
6789 @findex cdr, @r{introduced}
6791 In Lisp, @code{car}, @code{cdr}, and @code{cons} are fundamental
6792 functions. The @code{cons} function is used to construct lists, and
6793 the @code{car} and @code{cdr} functions are used to take them apart.
6795 In the walk through of the @code{copy-region-as-kill} function, we
6796 will see @code{cons} as well as two variants on @code{cdr},
6797 namely, @code{setcdr} and @code{nthcdr}. (@xref{copy-region-as-kill}.)
6800 * Strange Names:: An historical aside: why the strange names?
6801 * car & cdr:: Functions for extracting part of a list.
6802 * cons:: Constructing a list.
6803 * nthcdr:: Calling @code{cdr} repeatedly.
6805 * setcar:: Changing the first element of a list.
6806 * setcdr:: Changing the rest of a list.
6812 @unnumberedsec Strange Names
6815 The name of the @code{cons} function is not unreasonable: it is an
6816 abbreviation of the word `construct'. The origins of the names for
6817 @code{car} and @code{cdr}, on the other hand, are esoteric: @code{car}
6818 is an acronym from the phrase `Contents of the Address part of the
6819 Register'; and @code{cdr} (pronounced `could-er') is an acronym from
6820 the phrase `Contents of the Decrement part of the Register'. These
6821 phrases refer to specific pieces of hardware on the very early
6822 computer on which the original Lisp was developed. Besides being
6823 obsolete, the phrases have been completely irrelevant for more than 25
6824 years to anyone thinking about Lisp. Nonetheless, although a few
6825 brave scholars have begun to use more reasonable names for these
6826 functions, the old terms are still in use. In particular, since the
6827 terms are used in the Emacs Lisp source code, we will use them in this
6831 @section @code{car} and @code{cdr}
6833 The @sc{car} of a list is, quite simply, the first item in the list.
6834 Thus the @sc{car} of the list @code{(rose violet daisy buttercup)} is
6838 If you are reading this in Info in GNU Emacs, you can see this by
6839 evaluating the following:
6842 (car '(rose violet daisy buttercup))
6846 After evaluating the expression, @code{rose} will appear in the echo
6849 Clearly, a more reasonable name for the @code{car} function would be
6850 @code{first} and this is often suggested.
6852 @code{car} does not remove the first item from the list; it only reports
6853 what it is. After @code{car} has been applied to a list, the list is
6854 still the same as it was. In the jargon, @code{car} is
6855 `non-destructive'. This feature turns out to be important.
6857 The @sc{cdr} of a list is the rest of the list, that is, the
6858 @code{cdr} function returns the part of the list that follows the
6859 first item. Thus, while the @sc{car} of the list @code{'(rose violet
6860 daisy buttercup)} is @code{rose}, the rest of the list, the value
6861 returned by the @code{cdr} function, is @code{(violet daisy
6865 You can see this by evaluating the following in the usual way:
6868 (cdr '(rose violet daisy buttercup))
6872 When you evaluate this, @code{(violet daisy buttercup)} will appear in
6875 Like @code{car}, @code{cdr} does not remove any elements from the
6876 list---it just returns a report of what the second and subsequent
6879 Incidentally, in the example, the list of flowers is quoted. If it were
6880 not, the Lisp interpreter would try to evaluate the list by calling
6881 @code{rose} as a function. In this example, we do not want to do that.
6883 Clearly, a more reasonable name for @code{cdr} would be @code{rest}.
6885 (There is a lesson here: when you name new functions, consider very
6886 carefully what you are doing, since you may be stuck with the names
6887 for far longer than you expect. The reason this document perpetuates
6888 these names is that the Emacs Lisp source code uses them, and if I did
6889 not use them, you would have a hard time reading the code; but do,
6890 please, try to avoid using these terms yourself. The people who come
6891 after you will be grateful to you.)
6893 When @code{car} and @code{cdr} are applied to a list made up of symbols,
6894 such as the list @code{(pine fir oak maple)}, the element of the list
6895 returned by the function @code{car} is the symbol @code{pine} without
6896 any parentheses around it. @code{pine} is the first element in the
6897 list. However, the @sc{cdr} of the list is a list itself, @code{(fir
6898 oak maple)}, as you can see by evaluating the following expressions in
6903 (car '(pine fir oak maple))
6905 (cdr '(pine fir oak maple))
6909 On the other hand, in a list of lists, the first element is itself a
6910 list. @code{car} returns this first element as a list. For example,
6911 the following list contains three sub-lists, a list of carnivores, a
6912 list of herbivores and a list of sea mammals:
6916 (car '((lion tiger cheetah)
6917 (gazelle antelope zebra)
6918 (whale dolphin seal)))
6923 In this example, the first element or @sc{car} of the list is the list of
6924 carnivores, @code{(lion tiger cheetah)}, and the rest of the list is
6925 @code{((gazelle antelope zebra) (whale dolphin seal))}.
6929 (cdr '((lion tiger cheetah)
6930 (gazelle antelope zebra)
6931 (whale dolphin seal)))
6935 It is worth saying again that @code{car} and @code{cdr} are
6936 non-destructive---that is, they do not modify or change lists to which
6937 they are applied. This is very important for how they are used.
6939 Also, in the first chapter, in the discussion about atoms, I said that
6940 in Lisp, ``certain kinds of atom, such as an array, can be separated
6941 into parts; but the mechanism for doing this is different from the
6942 mechanism for splitting a list. As far as Lisp is concerned, the
6943 atoms of a list are unsplittable.'' (@xref{Lisp Atoms}.) The
6944 @code{car} and @code{cdr} functions are used for splitting lists and
6945 are considered fundamental to Lisp. Since they cannot split or gain
6946 access to the parts of an array, an array is considered an atom.
6947 Conversely, the other fundamental function, @code{cons}, can put
6948 together or construct a list, but not an array. (Arrays are handled
6949 by array-specific functions. @xref{Arrays, , Arrays, elisp, The GNU
6950 Emacs Lisp Reference Manual}.)
6953 @section @code{cons}
6954 @findex cons, @r{introduced}
6956 The @code{cons} function constructs lists; it is the inverse of
6957 @code{car} and @code{cdr}. For example, @code{cons} can be used to make
6958 a four element list from the three element list, @code{(fir oak maple)}:
6961 (cons 'pine '(fir oak maple))
6966 After evaluating this list, you will see
6969 (pine fir oak maple)
6973 appear in the echo area. @code{cons} causes the creation of a new
6974 list in which the element is followed by the elements of the original
6977 We often say that `@code{cons} puts a new element at the beginning of
6978 a list; it attaches or pushes elements onto the list', but this
6979 phrasing can be misleading, since @code{cons} does not change an
6980 existing list, but creates a new one.
6982 Like @code{car} and @code{cdr}, @code{cons} is non-destructive.
6986 * length:: How to find the length of a list.
6991 @unnumberedsubsec Build a list
6994 @code{cons} must have a list to attach to.@footnote{Actually, you can
6995 @code{cons} an element to an atom to produce a dotted pair. Dotted
6996 pairs are not discussed here; see @ref{Dotted Pair Notation, , Dotted
6997 Pair Notation, elisp, The GNU Emacs Lisp Reference Manual}.} You
6998 cannot start from absolutely nothing. If you are building a list, you
6999 need to provide at least an empty list at the beginning. Here is a
7000 series of @code{cons} expressions that build up a list of flowers. If
7001 you are reading this in Info in GNU Emacs, you can evaluate each of
7002 the expressions in the usual way; the value is printed in this text
7003 after @samp{@result{}}, which you may read as `evaluates to'.
7007 (cons 'buttercup ())
7008 @result{} (buttercup)
7012 (cons 'daisy '(buttercup))
7013 @result{} (daisy buttercup)
7017 (cons 'violet '(daisy buttercup))
7018 @result{} (violet daisy buttercup)
7022 (cons 'rose '(violet daisy buttercup))
7023 @result{} (rose violet daisy buttercup)
7028 In the first example, the empty list is shown as @code{()} and a list
7029 made up of @code{buttercup} followed by the empty list is constructed.
7030 As you can see, the empty list is not shown in the list that was
7031 constructed. All that you see is @code{(buttercup)}. The empty list is
7032 not counted as an element of a list because there is nothing in an empty
7033 list. Generally speaking, an empty list is invisible.
7035 The second example, @code{(cons 'daisy '(buttercup))} constructs a new,
7036 two element list by putting @code{daisy} in front of @code{buttercup};
7037 and the third example constructs a three element list by putting
7038 @code{violet} in front of @code{daisy} and @code{buttercup}.
7041 @subsection Find the Length of a List: @code{length}
7044 You can find out how many elements there are in a list by using the Lisp
7045 function @code{length}, as in the following examples:
7049 (length '(buttercup))
7054 (length '(daisy buttercup))
7059 (length (cons 'violet '(daisy buttercup)))
7065 In the third example, the @code{cons} function is used to construct a
7066 three element list which is then passed to the @code{length} function as
7070 We can also use @code{length} to count the number of elements in an
7081 As you would expect, the number of elements in an empty list is zero.
7083 An interesting experiment is to find out what happens if you try to find
7084 the length of no list at all; that is, if you try to call @code{length}
7085 without giving it an argument, not even an empty list:
7093 What you see, if you evaluate this, is the error message
7096 Lisp error: (wrong-number-of-arguments length 0)
7100 This means that the function receives the wrong number of
7101 arguments, zero, when it expects some other number of arguments. In
7102 this case, one argument is expected, the argument being a list whose
7103 length the function is measuring. (Note that @emph{one} list is
7104 @emph{one} argument, even if the list has many elements inside it.)
7106 The part of the error message that says @samp{length} is the name of
7110 @code{length} is still a subroutine, but you need C-h f to discover that.
7112 In an earlier version:
7113 This is written with a special notation, @samp{#<subr},
7114 that indicates that the function @code{length} is one of the primitive
7115 functions written in C rather than in Emacs Lisp. (@samp{subr} is an
7116 abbreviation for `subroutine'.) @xref{What Is a Function, , What Is a
7117 Function?, elisp , The GNU Emacs Lisp Reference Manual}, for more
7122 @section @code{nthcdr}
7125 The @code{nthcdr} function is associated with the @code{cdr} function.
7126 What it does is take the @sc{cdr} of a list repeatedly.
7128 If you take the @sc{cdr} of the list @code{(pine fir
7129 oak maple)}, you will be returned the list @code{(fir oak maple)}. If you
7130 repeat this on what was returned, you will be returned the list
7131 @code{(oak maple)}. (Of course, repeated @sc{cdr}ing on the original
7132 list will just give you the original @sc{cdr} since the function does
7133 not change the list. You need to evaluate the @sc{cdr} of the
7134 @sc{cdr} and so on.) If you continue this, eventually you will be
7135 returned an empty list, which in this case, instead of being shown as
7136 @code{()} is shown as @code{nil}.
7139 For review, here is a series of repeated @sc{cdr}s, the text following
7140 the @samp{@result{}} shows what is returned.
7144 (cdr '(pine fir oak maple))
7145 @result{}(fir oak maple)
7149 (cdr '(fir oak maple))
7150 @result{} (oak maple)
7175 You can also do several @sc{cdr}s without printing the values in
7180 (cdr (cdr '(pine fir oak maple)))
7181 @result{} (oak maple)
7186 In this example, the Lisp interpreter evaluates the innermost list first.
7187 The innermost list is quoted, so it just passes the list as it is to the
7188 innermost @code{cdr}. This @code{cdr} passes a list made up of the
7189 second and subsequent elements of the list to the outermost @code{cdr},
7190 which produces a list composed of the third and subsequent elements of
7191 the original list. In this example, the @code{cdr} function is repeated
7192 and returns a list that consists of the original list without its
7195 The @code{nthcdr} function does the same as repeating the call to
7196 @code{cdr}. In the following example, the argument 2 is passed to the
7197 function @code{nthcdr}, along with the list, and the value returned is
7198 the list without its first two items, which is exactly the same
7199 as repeating @code{cdr} twice on the list:
7203 (nthcdr 2 '(pine fir oak maple))
7204 @result{} (oak maple)
7209 Using the original four element list, we can see what happens when
7210 various numeric arguments are passed to @code{nthcdr}, including 0, 1,
7215 ;; @r{Leave the list as it was.}
7216 (nthcdr 0 '(pine fir oak maple))
7217 @result{} (pine fir oak maple)
7221 ;; @r{Return a copy without the first element.}
7222 (nthcdr 1 '(pine fir oak maple))
7223 @result{} (fir oak maple)
7227 ;; @r{Return a copy of the list without three elements.}
7228 (nthcdr 3 '(pine fir oak maple))
7233 ;; @r{Return a copy lacking all four elements.}
7234 (nthcdr 4 '(pine fir oak maple))
7239 ;; @r{Return a copy lacking all elements.}
7240 (nthcdr 5 '(pine fir oak maple))
7249 The @code{nthcdr} function takes the @sc{cdr} of a list repeatedly.
7250 The @code{nth} function takes the @sc{car} of the result returned by
7251 @code{nthcdr}. It returns the Nth element of the list.
7254 Thus, if it were not defined in C for speed, the definition of
7255 @code{nth} would be:
7260 "Returns the Nth element of LIST.
7261 N counts from zero. If LIST is not that long, nil is returned."
7262 (car (nthcdr n list)))
7267 (Originally, @code{nth} was defined in Emacs Lisp in @file{subr.el},
7268 but its definition was redone in C in the 1980s.)
7270 The @code{nth} function returns a single element of a list.
7271 This can be very convenient.
7273 Note that the elements are numbered from zero, not one. That is to
7274 say, the first element of a list, its @sc{car} is the zeroth element.
7275 This is called `zero-based' counting and often bothers people who
7276 are accustomed to the first element in a list being number one, which
7284 (nth 0 '("one" "two" "three"))
7287 (nth 1 '("one" "two" "three"))
7292 It is worth mentioning that @code{nth}, like @code{nthcdr} and
7293 @code{cdr}, does not change the original list---the function is
7294 non-destructive. This is in sharp contrast to the @code{setcar} and
7295 @code{setcdr} functions.
7298 @section @code{setcar}
7301 As you might guess from their names, the @code{setcar} and @code{setcdr}
7302 functions set the @sc{car} or the @sc{cdr} of a list to a new value.
7303 They actually change the original list, unlike @code{car} and @code{cdr}
7304 which leave the original list as it was. One way to find out how this
7305 works is to experiment. We will start with the @code{setcar} function.
7308 First, we can make a list and then set the value of a variable to the
7309 list, using the @code{setq} function. Here is a list of animals:
7312 (setq animals '(antelope giraffe lion tiger))
7316 If you are reading this in Info inside of GNU Emacs, you can evaluate
7317 this expression in the usual fashion, by positioning the cursor after
7318 the expression and typing @kbd{C-x C-e}. (I'm doing this right here
7319 as I write this. This is one of the advantages of having the
7320 interpreter built into the computing environment. Incidentally, when
7321 there is nothing on the line after the final parentheses, such as a
7322 comment, point can be on the next line. Thus, if your cursor is in
7323 the first column of the next line, you do not need to move it.
7324 Indeed, Emacs permits any amount of white space after the final
7328 When we evaluate the variable @code{animals}, we see that it is bound to
7329 the list @code{(antelope giraffe lion tiger)}:
7334 @result{} (antelope giraffe lion tiger)
7339 Put another way, the variable @code{animals} points to the list
7340 @code{(antelope giraffe lion tiger)}.
7342 Next, evaluate the function @code{setcar} while passing it two
7343 arguments, the variable @code{animals} and the quoted symbol
7344 @code{hippopotamus}; this is done by writing the three element list
7345 @code{(setcar animals 'hippopotamus)} and then evaluating it in the
7349 (setcar animals 'hippopotamus)
7354 After evaluating this expression, evaluate the variable @code{animals}
7355 again. You will see that the list of animals has changed:
7360 @result{} (hippopotamus giraffe lion tiger)
7365 The first element on the list, @code{antelope} is replaced by
7366 @code{hippopotamus}.
7368 So we can see that @code{setcar} did not add a new element to the list
7369 as @code{cons} would have; it replaced @code{antelope} with
7370 @code{hippopotamus}; it @emph{changed} the list.
7373 @section @code{setcdr}
7376 The @code{setcdr} function is similar to the @code{setcar} function,
7377 except that the function replaces the second and subsequent elements of
7378 a list rather than the first element.
7380 (To see how to change the last element of a list, look ahead to
7381 @ref{kill-new function, , The @code{kill-new} function}, which uses
7382 the @code{nthcdr} and @code{setcdr} functions.)
7385 To see how this works, set the value of the variable to a list of
7386 domesticated animals by evaluating the following expression:
7389 (setq domesticated-animals '(horse cow sheep goat))
7394 If you now evaluate the list, you will be returned the list
7395 @code{(horse cow sheep goat)}:
7399 domesticated-animals
7400 @result{} (horse cow sheep goat)
7405 Next, evaluate @code{setcdr} with two arguments, the name of the
7406 variable which has a list as its value, and the list to which the
7407 @sc{cdr} of the first list will be set;
7410 (setcdr domesticated-animals '(cat dog))
7414 If you evaluate this expression, the list @code{(cat dog)} will appear
7415 in the echo area. This is the value returned by the function. The
7416 result we are interested in is the ``side effect'', which we can see by
7417 evaluating the variable @code{domesticated-animals}:
7421 domesticated-animals
7422 @result{} (horse cat dog)
7427 Indeed, the list is changed from @code{(horse cow sheep goat)} to
7428 @code{(horse cat dog)}. The @sc{cdr} of the list is changed from
7429 @code{(cow sheep goat)} to @code{(cat dog)}.
7434 Construct a list of four birds by evaluating several expressions with
7435 @code{cons}. Find out what happens when you @code{cons} a list onto
7436 itself. Replace the first element of the list of four birds with a
7437 fish. Replace the rest of that list with a list of other fish.
7439 @node Cutting & Storing Text
7440 @chapter Cutting and Storing Text
7441 @cindex Cutting and storing text
7442 @cindex Storing and cutting text
7443 @cindex Killing text
7444 @cindex Clipping text
7445 @cindex Erasing text
7446 @cindex Deleting text
7448 Whenever you cut or clip text out of a buffer with a `kill' command in
7449 GNU Emacs, it is stored in a list and you can bring it back with a
7452 (The use of the word `kill' in Emacs for processes which specifically
7453 @emph{do not} destroy the values of the entities is an unfortunate
7454 historical accident. A much more appropriate word would be `clip' since
7455 that is what the kill commands do; they clip text out of a buffer and
7456 put it into storage from which it can be brought back. I have often
7457 been tempted to replace globally all occurrences of `kill' in the Emacs
7458 sources with `clip' and all occurrences of `killed' with `clipped'.)
7461 * Storing Text:: Text is stored in a list.
7462 * zap-to-char:: Cutting out text up to a character.
7463 * kill-region:: Cutting text out of a region.
7464 * copy-region-as-kill:: A definition for copying text.
7465 * Digression into C:: Minor note on C programming language macros.
7466 * defvar:: How to give a variable an initial value.
7467 * cons & search-fwd Review::
7468 * search Exercises::
7473 @unnumberedsec Storing Text in a List
7476 When text is cut out of a buffer, it is stored on a list. Successive
7477 pieces of text are stored on the list successively, so the list might
7481 ("a piece of text" "previous piece")
7486 The function @code{cons} can be used to create a new list from a piece
7487 of text (an `atom', to use the jargon) and an existing list, like
7492 (cons "another piece"
7493 '("a piece of text" "previous piece"))
7499 If you evaluate this expression, a list of three elements will appear in
7503 ("another piece" "a piece of text" "previous piece")
7506 With the @code{car} and @code{nthcdr} functions, you can retrieve
7507 whichever piece of text you want. For example, in the following code,
7508 @code{nthcdr 1 @dots{}} returns the list with the first item removed;
7509 and the @code{car} returns the first element of that remainder---the
7510 second element of the original list:
7514 (car (nthcdr 1 '("another piece"
7517 @result{} "a piece of text"
7521 The actual functions in Emacs are more complex than this, of course.
7522 The code for cutting and retrieving text has to be written so that
7523 Emacs can figure out which element in the list you want---the first,
7524 second, third, or whatever. In addition, when you get to the end of
7525 the list, Emacs should give you the first element of the list, rather
7526 than nothing at all.
7528 The list that holds the pieces of text is called the @dfn{kill ring}.
7529 This chapter leads up to a description of the kill ring and how it is
7530 used by first tracing how the @code{zap-to-char} function works. This
7531 function uses (or `calls') a function that invokes a function that
7532 manipulates the kill ring. Thus, before reaching the mountains, we
7533 climb the foothills.
7535 A subsequent chapter describes how text that is cut from the buffer is
7536 retrieved. @xref{Yanking, , Yanking Text Back}.
7539 @section @code{zap-to-char}
7542 Let us look at the interactive @code{zap-to-char} function.
7545 * Complete zap-to-char:: The complete implementation.
7546 * zap-to-char interactive:: A three part interactive expression.
7547 * zap-to-char body:: A short overview.
7548 * search-forward:: How to search for a string.
7549 * progn:: The @code{progn} special form.
7550 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
7554 @node Complete zap-to-char
7555 @unnumberedsubsec The Complete @code{zap-to-char} Implementation
7558 The @code{zap-to-char} function removes the text in the region between
7559 the location of the cursor (i.e., of point) up to and including the
7560 next occurrence of a specified character. The text that
7561 @code{zap-to-char} removes is put in the kill ring; and it can be
7562 retrieved from the kill ring by typing @kbd{C-y} (@code{yank}). If
7563 the command is given an argument, it removes text through that number
7564 of occurrences. Thus, if the cursor were at the beginning of this
7565 sentence and the character were @samp{s}, @samp{Thus} would be
7566 removed. If the argument were two, @samp{Thus, if the curs} would be
7567 removed, up to and including the @samp{s} in @samp{cursor}.
7569 If the specified character is not found, @code{zap-to-char} will say
7570 ``Search failed'', tell you the character you typed, and not remove
7573 In order to determine how much text to remove, @code{zap-to-char} uses
7574 a search function. Searches are used extensively in code that
7575 manipulates text, and we will focus attention on them as well as on the
7579 @c GNU Emacs version 19
7580 (defun zap-to-char (arg char) ; version 19 implementation
7581 "Kill up to and including ARG'th occurrence of CHAR.
7582 Goes backward if ARG is negative; error if CHAR not found."
7583 (interactive "*p\ncZap to char: ")
7584 (kill-region (point)
7587 (char-to-string char) nil nil arg)
7592 Here is the complete text of the version 22 implementation of the function:
7597 (defun zap-to-char (arg char)
7598 "Kill up to and including ARG'th occurrence of CHAR.
7599 Case is ignored if `case-fold-search' is non-nil in the current buffer.
7600 Goes backward if ARG is negative; error if CHAR not found."
7601 (interactive "p\ncZap to char: ")
7602 (if (char-table-p translation-table-for-input)
7603 (setq char (or (aref translation-table-for-input char) char)))
7604 (kill-region (point) (progn
7605 (search-forward (char-to-string char)
7611 The documentation is thorough. You do need to know the jargon meaning
7614 @node zap-to-char interactive
7615 @subsection The @code{interactive} Expression
7618 The interactive expression in the @code{zap-to-char} command looks like
7622 (interactive "p\ncZap to char: ")
7625 The part within quotation marks, @code{"p\ncZap to char:@: "}, specifies
7626 two different things. First, and most simply, is the @samp{p}.
7627 This part is separated from the next part by a newline, @samp{\n}.
7628 The @samp{p} means that the first argument to the function will be
7629 passed the value of a `processed prefix'. The prefix argument is
7630 passed by typing @kbd{C-u} and a number, or @kbd{M-} and a number. If
7631 the function is called interactively without a prefix, 1 is passed to
7634 The second part of @code{"p\ncZap to char:@: "} is
7635 @samp{cZap to char:@: }. In this part, the lower case @samp{c}
7636 indicates that @code{interactive} expects a prompt and that the
7637 argument will be a character. The prompt follows the @samp{c} and is
7638 the string @samp{Zap to char:@: } (with a space after the colon to
7641 What all this does is prepare the arguments to @code{zap-to-char} so they
7642 are of the right type, and give the user a prompt.
7644 In a read-only buffer, the @code{zap-to-char} function copies the text
7645 to the kill ring, but does not remove it. The echo area displays a
7646 message saying that the buffer is read-only. Also, the terminal may
7647 beep or blink at you.
7649 @node zap-to-char body
7650 @subsection The Body of @code{zap-to-char}
7652 The body of the @code{zap-to-char} function contains the code that
7653 kills (that is, removes) the text in the region from the current
7654 position of the cursor up to and including the specified character.
7656 The first part of the code looks like this:
7659 (if (char-table-p translation-table-for-input)
7660 (setq char (or (aref translation-table-for-input char) char)))
7661 (kill-region (point) (progn
7662 (search-forward (char-to-string char) nil nil arg)
7667 @code{char-table-p} is an hitherto unseen function. It determines
7668 whether its argument is a character table. When it is, it sets the
7669 character passed to @code{zap-to-char} to one of them, if that
7670 character exists, or to the character itself. (This becomes important
7671 for certain characters in non-European languages. The @code{aref}
7672 function extracts an element from an array. It is an array-specific
7673 function that is not described in this document. @xref{Arrays, ,
7674 Arrays, elisp, The GNU Emacs Lisp Reference Manual}.)
7677 @code{(point)} is the current position of the cursor.
7679 The next part of the code is an expression using @code{progn}. The body
7680 of the @code{progn} consists of calls to @code{search-forward} and
7683 It is easier to understand how @code{progn} works after learning about
7684 @code{search-forward}, so we will look at @code{search-forward} and
7685 then at @code{progn}.
7687 @node search-forward
7688 @subsection The @code{search-forward} Function
7689 @findex search-forward
7691 The @code{search-forward} function is used to locate the
7692 zapped-for-character in @code{zap-to-char}. If the search is
7693 successful, @code{search-forward} leaves point immediately after the
7694 last character in the target string. (In @code{zap-to-char}, the
7695 target string is just one character long. @code{zap-to-char} uses the
7696 function @code{char-to-string} to ensure that the computer treats that
7697 character as a string.) If the search is backwards,
7698 @code{search-forward} leaves point just before the first character in
7699 the target. Also, @code{search-forward} returns @code{t} for true.
7700 (Moving point is therefore a `side effect'.)
7703 In @code{zap-to-char}, the @code{search-forward} function looks like this:
7706 (search-forward (char-to-string char) nil nil arg)
7709 The @code{search-forward} function takes four arguments:
7713 The first argument is the target, what is searched for. This must be a
7714 string, such as @samp{"z"}.
7716 As it happens, the argument passed to @code{zap-to-char} is a single
7717 character. Because of the way computers are built, the Lisp
7718 interpreter may treat a single character as being different from a
7719 string of characters. Inside the computer, a single character has a
7720 different electronic format than a string of one character. (A single
7721 character can often be recorded in the computer using exactly one
7722 byte; but a string may be longer, and the computer needs to be ready
7723 for this.) Since the @code{search-forward} function searches for a
7724 string, the character that the @code{zap-to-char} function receives as
7725 its argument must be converted inside the computer from one format to
7726 the other; otherwise the @code{search-forward} function will fail.
7727 The @code{char-to-string} function is used to make this conversion.
7730 The second argument bounds the search; it is specified as a position in
7731 the buffer. In this case, the search can go to the end of the buffer,
7732 so no bound is set and the second argument is @code{nil}.
7735 The third argument tells the function what it should do if the search
7736 fails---it can signal an error (and print a message) or it can return
7737 @code{nil}. A @code{nil} as the third argument causes the function to
7738 signal an error when the search fails.
7741 The fourth argument to @code{search-forward} is the repeat count---how
7742 many occurrences of the string to look for. This argument is optional
7743 and if the function is called without a repeat count, this argument is
7744 passed the value 1. If this argument is negative, the search goes
7749 In template form, a @code{search-forward} expression looks like this:
7753 (search-forward "@var{target-string}"
7754 @var{limit-of-search}
7755 @var{what-to-do-if-search-fails}
7760 We will look at @code{progn} next.
7763 @subsection The @code{progn} Special Form
7766 @code{progn} is a special form that causes each of its arguments to be
7767 evaluated in sequence and then returns the value of the last one. The
7768 preceding expressions are evaluated only for the side effects they
7769 perform. The values produced by them are discarded.
7772 The template for a @code{progn} expression is very simple:
7781 In @code{zap-to-char}, the @code{progn} expression has to do two things:
7782 put point in exactly the right position; and return the location of
7783 point so that @code{kill-region} will know how far to kill to.
7785 The first argument to the @code{progn} is @code{search-forward}. When
7786 @code{search-forward} finds the string, the function leaves point
7787 immediately after the last character in the target string. (In this
7788 case the target string is just one character long.) If the search is
7789 backwards, @code{search-forward} leaves point just before the first
7790 character in the target. The movement of point is a side effect.
7792 The second and last argument to @code{progn} is the expression
7793 @code{(point)}. This expression returns the value of point, which in
7794 this case will be the location to which it has been moved by
7795 @code{search-forward}. (In the source, a line that tells the function
7796 to go to the previous character, if it is going forward, was commented
7797 out in 1999; I don't remember whether that feature or mis-feature was
7798 ever a part of the distributed source.) The value of @code{point} is
7799 returned by the @code{progn} expression and is passed to
7800 @code{kill-region} as @code{kill-region}'s second argument.
7802 @node Summing up zap-to-char
7803 @subsection Summing up @code{zap-to-char}
7805 Now that we have seen how @code{search-forward} and @code{progn} work,
7806 we can see how the @code{zap-to-char} function works as a whole.
7808 The first argument to @code{kill-region} is the position of the cursor
7809 when the @code{zap-to-char} command is given---the value of point at
7810 that time. Within the @code{progn}, the search function then moves
7811 point to just after the zapped-to-character and @code{point} returns the
7812 value of this location. The @code{kill-region} function puts together
7813 these two values of point, the first one as the beginning of the region
7814 and the second one as the end of the region, and removes the region.
7816 The @code{progn} special form is necessary because the
7817 @code{kill-region} command takes two arguments; and it would fail if
7818 @code{search-forward} and @code{point} expressions were written in
7819 sequence as two additional arguments. The @code{progn} expression is
7820 a single argument to @code{kill-region} and returns the one value that
7821 @code{kill-region} needs for its second argument.
7824 @section @code{kill-region}
7827 The @code{zap-to-char} function uses the @code{kill-region} function.
7828 This function clips text from a region and copies that text to
7829 the kill ring, from which it may be retrieved.
7834 (defun kill-region (beg end &optional yank-handler)
7835 "Kill (\"cut\") text between point and mark.
7836 This deletes the text from the buffer and saves it in the kill ring.
7837 The command \\[yank] can retrieve it from there.
7838 \(If you want to kill and then yank immediately, use \\[kill-ring-save].)
7840 If you want to append the killed region to the last killed text,
7841 use \\[append-next-kill] before \\[kill-region].
7843 If the buffer is read-only, Emacs will beep and refrain from deleting
7844 the text, but put the text in the kill ring anyway. This means that
7845 you can use the killing commands to copy text from a read-only buffer.
7847 This is the primitive for programs to kill text (as opposed to deleting it).
7848 Supply two arguments, character positions indicating the stretch of text
7850 Any command that calls this function is a \"kill command\".
7851 If the previous command was also a kill command,
7852 the text killed this time appends to the text killed last time
7853 to make one entry in the kill ring.
7855 In Lisp code, optional third arg YANK-HANDLER, if non-nil,
7856 specifies the yank-handler text property to be set on the killed
7857 text. See `insert-for-yank'."
7858 ;; Pass point first, then mark, because the order matters
7859 ;; when calling kill-append.
7860 (interactive (list (point) (mark)))
7861 (unless (and beg end)
7862 (error "The mark is not set now, so there is no region"))
7864 (let ((string (filter-buffer-substring beg end t)))
7865 (when string ;STRING is nil if BEG = END
7866 ;; Add that string to the kill ring, one way or another.
7867 (if (eq last-command 'kill-region)
7868 (kill-append string (< end beg) yank-handler)
7869 (kill-new string nil yank-handler)))
7870 (when (or string (eq last-command 'kill-region))
7871 (setq this-command 'kill-region))
7873 ((buffer-read-only text-read-only)
7874 ;; The code above failed because the buffer, or some of the characters
7875 ;; in the region, are read-only.
7876 ;; We should beep, in case the user just isn't aware of this.
7877 ;; However, there's no harm in putting
7878 ;; the region's text in the kill ring, anyway.
7879 (copy-region-as-kill beg end)
7880 ;; Set this-command now, so it will be set even if we get an error.
7881 (setq this-command 'kill-region)
7882 ;; This should barf, if appropriate, and give us the correct error.
7883 (if kill-read-only-ok
7884 (progn (message "Read only text copied to kill ring") nil)
7885 ;; Signal an error if the buffer is read-only.
7886 (barf-if-buffer-read-only)
7887 ;; If the buffer isn't read-only, the text is.
7888 (signal 'text-read-only (list (current-buffer)))))))
7891 The Emacs 22 version of that function uses @code{condition-case} and
7892 @code{copy-region-as-kill}, both of which we will explain.
7893 @code{condition-case} is an important special form.
7895 In essence, the @code{kill-region} function calls
7896 @code{condition-case}, which takes three arguments. In this function,
7897 the first argument does nothing. The second argument contains the
7898 code that does the work when all goes well. The third argument
7899 contains the code that is called in the event of an error.
7902 * Complete kill-region:: The function definition.
7903 * condition-case:: Dealing with a problem.
7908 @node Complete kill-region
7909 @unnumberedsubsec The Complete @code{kill-region} Definition
7913 We will go through the @code{condition-case} code in a moment. First,
7914 let us look at the definition of @code{kill-region}, with comments
7920 (defun kill-region (beg end)
7921 "Kill (\"cut\") text between point and mark.
7922 This deletes the text from the buffer and saves it in the kill ring.
7923 The command \\[yank] can retrieve it from there. @dots{} "
7927 ;; @bullet{} Since order matters, pass point first.
7928 (interactive (list (point) (mark)))
7929 ;; @bullet{} And tell us if we cannot cut the text.
7930 ;; `unless' is an `if' without a then-part.
7931 (unless (and beg end)
7932 (error "The mark is not set now, so there is no region"))
7936 ;; @bullet{} `condition-case' takes three arguments.
7937 ;; If the first argument is nil, as it is here,
7938 ;; information about the error signal is not
7939 ;; stored for use by another function.
7944 ;; @bullet{} The second argument to `condition-case' tells the
7945 ;; Lisp interpreter what to do when all goes well.
7949 ;; It starts with a `let' function that extracts the string
7950 ;; and tests whether it exists. If so (that is what the
7951 ;; `when' checks), it calls an `if' function that determines
7952 ;; whether the previous command was another call to
7953 ;; `kill-region'; if it was, then the new text is appended to
7954 ;; the previous text; if not, then a different function,
7955 ;; `kill-new', is called.
7959 ;; The `kill-append' function concatenates the new string and
7960 ;; the old. The `kill-new' function inserts text into a new
7961 ;; item in the kill ring.
7965 ;; `when' is an `if' without an else-part. The second `when'
7966 ;; again checks whether the current string exists; in
7967 ;; addition, it checks whether the previous command was
7968 ;; another call to `kill-region'. If one or the other
7969 ;; condition is true, then it sets the current command to
7970 ;; be `kill-region'.
7973 (let ((string (filter-buffer-substring beg end t)))
7974 (when string ;STRING is nil if BEG = END
7975 ;; Add that string to the kill ring, one way or another.
7976 (if (eq last-command 'kill-region)
7979 ;; @minus{} `yank-handler' is an optional argument to
7980 ;; `kill-region' that tells the `kill-append' and
7981 ;; `kill-new' functions how deal with properties
7982 ;; added to the text, such as `bold' or `italics'.
7983 (kill-append string (< end beg) yank-handler)
7984 (kill-new string nil yank-handler)))
7985 (when (or string (eq last-command 'kill-region))
7986 (setq this-command 'kill-region))
7991 ;; @bullet{} The third argument to `condition-case' tells the interpreter
7992 ;; what to do with an error.
7995 ;; The third argument has a conditions part and a body part.
7996 ;; If the conditions are met (in this case,
7997 ;; if text or buffer are read-only)
7998 ;; then the body is executed.
8001 ;; The first part of the third argument is the following:
8002 ((buffer-read-only text-read-only) ;; the if-part
8003 ;; @dots{} the then-part
8004 (copy-region-as-kill beg end)
8007 ;; Next, also as part of the then-part, set this-command, so
8008 ;; it will be set in an error
8009 (setq this-command 'kill-region)
8010 ;; Finally, in the then-part, send a message if you may copy
8011 ;; the text to the kill ring without signaling an error, but
8012 ;; don't if you may not.
8015 (if kill-read-only-ok
8016 (progn (message "Read only text copied to kill ring") nil)
8017 (barf-if-buffer-read-only)
8018 ;; If the buffer isn't read-only, the text is.
8019 (signal 'text-read-only (list (current-buffer)))))
8027 (defun kill-region (beg end)
8028 "Kill between point and mark.
8029 The text is deleted but saved in the kill ring."
8034 ;; 1. `condition-case' takes three arguments.
8035 ;; If the first argument is nil, as it is here,
8036 ;; information about the error signal is not
8037 ;; stored for use by another function.
8042 ;; 2. The second argument to `condition-case'
8043 ;; tells the Lisp interpreter what to do when all goes well.
8047 ;; The `delete-and-extract-region' function usually does the
8048 ;; work. If the beginning and ending of the region are both
8049 ;; the same, then the variable `string' will be empty, or nil
8050 (let ((string (delete-and-extract-region beg end)))
8054 ;; `when' is an `if' clause that cannot take an `else-part'.
8055 ;; Emacs normally sets the value of `last-command' to the
8056 ;; previous command.
8059 ;; `kill-append' concatenates the new string and the old.
8060 ;; `kill-new' inserts text into a new item in the kill ring.
8062 (if (eq last-command 'kill-region)
8063 ;; if true, prepend string
8064 (kill-append string (< end beg))
8066 (setq this-command 'kill-region))
8070 ;; 3. The third argument to `condition-case' tells the interpreter
8071 ;; what to do with an error.
8074 ;; The third argument has a conditions part and a body part.
8075 ;; If the conditions are met (in this case,
8076 ;; if text or buffer are read-only)
8077 ;; then the body is executed.
8080 ((buffer-read-only text-read-only) ;; this is the if-part
8082 (copy-region-as-kill beg end)
8085 (if kill-read-only-ok ;; usually this variable is nil
8086 (message "Read only text copied to kill ring")
8087 ;; or else, signal an error if the buffer is read-only;
8088 (barf-if-buffer-read-only)
8089 ;; and, in any case, signal that the text is read-only.
8090 (signal 'text-read-only (list (current-buffer)))))))
8095 @node condition-case
8096 @subsection @code{condition-case}
8097 @findex condition-case
8099 As we have seen earlier (@pxref{Making Errors, , Generate an Error
8100 Message}), when the Emacs Lisp interpreter has trouble evaluating an
8101 expression, it provides you with help; in the jargon, this is called
8102 ``signaling an error''. Usually, the computer stops the program and
8103 shows you a message.
8105 However, some programs undertake complicated actions. They should not
8106 simply stop on an error. In the @code{kill-region} function, the most
8107 likely error is that you will try to kill text that is read-only and
8108 cannot be removed. So the @code{kill-region} function contains code
8109 to handle this circumstance. This code, which makes up the body of
8110 the @code{kill-region} function, is inside of a @code{condition-case}
8114 The template for @code{condition-case} looks like this:
8121 @var{error-handler}@dots{})
8125 The second argument, @var{bodyform}, is straightforward. The
8126 @code{condition-case} special form causes the Lisp interpreter to
8127 evaluate the code in @var{bodyform}. If no error occurs, the special
8128 form returns the code's value and produces the side-effects, if any.
8130 In short, the @var{bodyform} part of a @code{condition-case}
8131 expression determines what should happen when everything works
8134 However, if an error occurs, among its other actions, the function
8135 generating the error signal will define one or more error condition
8138 An error handler is the third argument to @code{condition case}.
8139 An error handler has two parts, a @var{condition-name} and a
8140 @var{body}. If the @var{condition-name} part of an error handler
8141 matches a condition name generated by an error, then the @var{body}
8142 part of the error handler is run.
8144 As you will expect, the @var{condition-name} part of an error handler
8145 may be either a single condition name or a list of condition names.
8147 Also, a complete @code{condition-case} expression may contain more
8148 than one error handler. When an error occurs, the first applicable
8151 Lastly, the first argument to the @code{condition-case} expression,
8152 the @var{var} argument, is sometimes bound to a variable that
8153 contains information about the error. However, if that argument is
8154 nil, as is the case in @code{kill-region}, that information is
8158 In brief, in the @code{kill-region} function, the code
8159 @code{condition-case} works like this:
8163 @var{If no errors}, @var{run only this code}
8164 @var{but}, @var{if errors}, @var{run this other code}.
8171 copy-region-as-kill is short, 12 lines, and uses
8172 filter-buffer-substring, which is longer, 39 lines
8173 and has delete-and-extract-region in it.
8174 delete-and-extract-region is written in C.
8176 see Initializing a Variable with @code{defvar}
8178 Initializing a Variable with @code{defvar} includes line 8350
8182 @subsection Lisp macro
8186 The part of the @code{condition-case} expression that is evaluated in
8187 the expectation that all goes well has a @code{when}. The code uses
8188 @code{when} to determine whether the @code{string} variable points to
8191 A @code{when} expression is simply a programmers' convenience. It is
8192 an @code{if} without the possibility of an else clause. In your mind,
8193 you can replace @code{when} with @code{if} and understand what goes
8194 on. That is what the Lisp interpreter does.
8196 Technically speaking, @code{when} is a Lisp macro. A Lisp macro
8197 enables you to define new control constructs and other language
8198 features. It tells the interpreter how to compute another Lisp
8199 expression which will in turn compute the value. In this case, the
8200 `other expression' is an @code{if} expression.
8202 The @code{kill-region} function definition also has an @code{unless}
8203 macro; it is the converse of @code{when}. The @code{unless} macro is
8204 an @code{if} without a then clause
8206 For more about Lisp macros, see @ref{Macros, , Macros, elisp, The GNU
8207 Emacs Lisp Reference Manual}. The C programming language also
8208 provides macros. These are different, but also useful.
8211 We will briefly look at C macros in
8212 @ref{Digression into C}.
8216 Regarding the @code{when} macro, in the @code{condition-case}
8217 expression, when the string has content, then another conditional
8218 expression is executed. This is an @code{if} with both a then-part
8223 (if (eq last-command 'kill-region)
8224 (kill-append string (< end beg) yank-handler)
8225 (kill-new string nil yank-handler))
8229 The then-part is evaluated if the previous command was another call to
8230 @code{kill-region}; if not, the else-part is evaluated.
8232 @code{yank-handler} is an optional argument to @code{kill-region} that
8233 tells the @code{kill-append} and @code{kill-new} functions how deal
8234 with properties added to the text, such as `bold' or `italics'.
8236 @code{last-command} is a variable that comes with Emacs that we have
8237 not seen before. Normally, whenever a function is executed, Emacs
8238 sets the value of @code{last-command} to the previous command.
8241 In this segment of the definition, the @code{if} expression checks
8242 whether the previous command was @code{kill-region}. If it was,
8245 (kill-append string (< end beg) yank-handler)
8249 concatenates a copy of the newly clipped text to the just previously
8250 clipped text in the kill ring.
8252 @node copy-region-as-kill
8253 @section @code{copy-region-as-kill}
8254 @findex copy-region-as-kill
8257 The @code{copy-region-as-kill} function copies a region of text from a
8258 buffer and (via either @code{kill-append} or @code{kill-new}) saves it
8259 in the @code{kill-ring}.
8261 If you call @code{copy-region-as-kill} immediately after a
8262 @code{kill-region} command, Emacs appends the newly copied text to the
8263 previously copied text. This means that if you yank back the text, you
8264 get it all, from both this and the previous operation. On the other
8265 hand, if some other command precedes the @code{copy-region-as-kill},
8266 the function copies the text into a separate entry in the kill ring.
8269 * Complete copy-region-as-kill:: The complete function definition.
8270 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
8274 @node Complete copy-region-as-kill
8275 @unnumberedsubsec The complete @code{copy-region-as-kill} function definition
8279 Here is the complete text of the version 22 @code{copy-region-as-kill}
8284 (defun copy-region-as-kill (beg end)
8285 "Save the region as if killed, but don't kill it.
8286 In Transient Mark mode, deactivate the mark.
8287 If `interprogram-cut-function' is non-nil, also save the text for a window
8288 system cut and paste."
8292 (if (eq last-command 'kill-region)
8293 (kill-append (filter-buffer-substring beg end) (< end beg))
8294 (kill-new (filter-buffer-substring beg end)))
8297 (if transient-mark-mode
8298 (setq deactivate-mark t))
8304 As usual, this function can be divided into its component parts:
8308 (defun copy-region-as-kill (@var{argument-list})
8309 "@var{documentation}@dots{}"
8315 The arguments are @code{beg} and @code{end} and the function is
8316 interactive with @code{"r"}, so the two arguments must refer to the
8317 beginning and end of the region. If you have been reading through this
8318 document from the beginning, understanding these parts of a function is
8319 almost becoming routine.
8321 The documentation is somewhat confusing unless you remember that the
8322 word `kill' has a meaning different from usual. The `Transient Mark'
8323 and @code{interprogram-cut-function} comments explain certain
8326 After you once set a mark, a buffer always contains a region. If you
8327 wish, you can use Transient Mark mode to highlight the region
8328 temporarily. (No one wants to highlight the region all the time, so
8329 Transient Mark mode highlights it only at appropriate times. Many
8330 people turn off Transient Mark mode, so the region is never
8333 Also, a windowing system allows you to copy, cut, and paste among
8334 different programs. In the X windowing system, for example, the
8335 @code{interprogram-cut-function} function is @code{x-select-text},
8336 which works with the windowing system's equivalent of the Emacs kill
8339 The body of the @code{copy-region-as-kill} function starts with an
8340 @code{if} clause. What this clause does is distinguish between two
8341 different situations: whether or not this command is executed
8342 immediately after a previous @code{kill-region} command. In the first
8343 case, the new region is appended to the previously copied text.
8344 Otherwise, it is inserted into the beginning of the kill ring as a
8345 separate piece of text from the previous piece.
8347 The last two lines of the function prevent the region from lighting up
8348 if Transient Mark mode is turned on.
8350 The body of @code{copy-region-as-kill} merits discussion in detail.
8352 @node copy-region-as-kill body
8353 @subsection The Body of @code{copy-region-as-kill}
8355 The @code{copy-region-as-kill} function works in much the same way as
8356 the @code{kill-region} function. Both are written so that two or more
8357 kills in a row combine their text into a single entry. If you yank
8358 back the text from the kill ring, you get it all in one piece.
8359 Moreover, kills that kill forward from the current position of the
8360 cursor are added to the end of the previously copied text and commands
8361 that copy text backwards add it to the beginning of the previously
8362 copied text. This way, the words in the text stay in the proper
8365 Like @code{kill-region}, the @code{copy-region-as-kill} function makes
8366 use of the @code{last-command} variable that keeps track of the
8367 previous Emacs command.
8370 * last-command & this-command::
8371 * kill-append function::
8372 * kill-new function::
8376 @node last-command & this-command
8377 @unnumberedsubsubsec @code{last-command} and @code{this-command}
8380 Normally, whenever a function is executed, Emacs sets the value of
8381 @code{this-command} to the function being executed (which in this case
8382 would be @code{copy-region-as-kill}). At the same time, Emacs sets
8383 the value of @code{last-command} to the previous value of
8384 @code{this-command}.
8386 In the first part of the body of the @code{copy-region-as-kill}
8387 function, an @code{if} expression determines whether the value of
8388 @code{last-command} is @code{kill-region}. If so, the then-part of
8389 the @code{if} expression is evaluated; it uses the @code{kill-append}
8390 function to concatenate the text copied at this call to the function
8391 with the text already in the first element (the @sc{car}) of the kill
8392 ring. On the other hand, if the value of @code{last-command} is not
8393 @code{kill-region}, then the @code{copy-region-as-kill} function
8394 attaches a new element to the kill ring using the @code{kill-new}
8398 The @code{if} expression reads as follows; it uses @code{eq}:
8402 (if (eq last-command 'kill-region)
8404 (kill-append (filter-buffer-substring beg end) (< end beg))
8406 (kill-new (filter-buffer-substring beg end)))
8410 @findex filter-buffer-substring
8411 (The @code{filter-buffer-substring} function returns a filtered
8412 substring of the buffer, if any. Optionally---the arguments are not
8413 here, so neither is done---the function may delete the initial text or
8414 return the text without its properties; this function is a replacement
8415 for the older @code{buffer-substring} function, which came before text
8416 properties were implemented.)
8418 @findex eq @r{(example of use)}
8420 The @code{eq} function tests whether its first argument is the same Lisp
8421 object as its second argument. The @code{eq} function is similar to the
8422 @code{equal} function in that it is used to test for equality, but
8423 differs in that it determines whether two representations are actually
8424 the same object inside the computer, but with different names.
8425 @code{equal} determines whether the structure and contents of two
8426 expressions are the same.
8428 If the previous command was @code{kill-region}, then the Emacs Lisp
8429 interpreter calls the @code{kill-append} function
8431 @node kill-append function
8432 @unnumberedsubsubsec The @code{kill-append} function
8436 The @code{kill-append} function looks like this:
8441 (defun kill-append (string before-p &optional yank-handler)
8442 "Append STRING to the end of the latest kill in the kill ring.
8443 If BEFORE-P is non-nil, prepend STRING to the kill.
8445 (let* ((cur (car kill-ring)))
8446 (kill-new (if before-p (concat string cur) (concat cur string))
8447 (or (= (length cur) 0)
8449 (get-text-property 0 'yank-handler cur)))
8456 (defun kill-append (string before-p)
8457 "Append STRING to the end of the latest kill in the kill ring.
8458 If BEFORE-P is non-nil, prepend STRING to the kill.
8459 If `interprogram-cut-function' is set, pass the resulting kill to
8461 (kill-new (if before-p
8462 (concat string (car kill-ring))
8463 (concat (car kill-ring) string))
8468 The @code{kill-append} function is fairly straightforward. It uses
8469 the @code{kill-new} function, which we will discuss in more detail in
8472 (Also, the function provides an optional argument called
8473 @code{yank-handler}; when invoked, this argument tells the function
8474 how to deal with properties added to the text, such as `bold' or
8477 @c !!! bug in GNU Emacs 22 version of kill-append ?
8478 It has a @code{let*} function to set the value of the first element of
8479 the kill ring to @code{cur}. (I do not know why the function does not
8480 use @code{let} instead; only one value is set in the expression.
8481 Perhaps this is a bug that produces no problems?)
8483 Consider the conditional that is one of the two arguments to
8484 @code{kill-new}. It uses @code{concat} to concatenate the new text to
8485 the @sc{car} of the kill ring. Whether it prepends or appends the
8486 text depends on the results of an @code{if} expression:
8490 (if before-p ; @r{if-part}
8491 (concat string cur) ; @r{then-part}
8492 (concat cur string)) ; @r{else-part}
8497 If the region being killed is before the region that was killed in the
8498 last command, then it should be prepended before the material that was
8499 saved in the previous kill; and conversely, if the killed text follows
8500 what was just killed, it should be appended after the previous text.
8501 The @code{if} expression depends on the predicate @code{before-p} to
8502 decide whether the newly saved text should be put before or after the
8503 previously saved text.
8505 The symbol @code{before-p} is the name of one of the arguments to
8506 @code{kill-append}. When the @code{kill-append} function is
8507 evaluated, it is bound to the value returned by evaluating the actual
8508 argument. In this case, this is the expression @code{(< end beg)}.
8509 This expression does not directly determine whether the killed text in
8510 this command is located before or after the kill text of the last
8511 command; what it does is determine whether the value of the variable
8512 @code{end} is less than the value of the variable @code{beg}. If it
8513 is, it means that the user is most likely heading towards the
8514 beginning of the buffer. Also, the result of evaluating the predicate
8515 expression, @code{(< end beg)}, will be true and the text will be
8516 prepended before the previous text. On the other hand, if the value of
8517 the variable @code{end} is greater than the value of the variable
8518 @code{beg}, the text will be appended after the previous text.
8521 When the newly saved text will be prepended, then the string with the new
8522 text will be concatenated before the old text:
8530 But if the text will be appended, it will be concatenated
8534 (concat cur string))
8537 To understand how this works, we first need to review the
8538 @code{concat} function. The @code{concat} function links together or
8539 unites two strings of text. The result is a string. For example:
8543 (concat "abc" "def")
8549 (car '("first element" "second element")))
8550 @result{} "new first element"
8553 '("first element" "second element")) " modified")
8554 @result{} "first element modified"
8558 We can now make sense of @code{kill-append}: it modifies the contents
8559 of the kill ring. The kill ring is a list, each element of which is
8560 saved text. The @code{kill-append} function uses the @code{kill-new}
8561 function which in turn uses the @code{setcar} function.
8563 @node kill-new function
8564 @unnumberedsubsubsec The @code{kill-new} function
8567 @c in GNU Emacs 22, additional documentation to kill-new:
8569 Optional third arguments YANK-HANDLER controls how the STRING is later
8570 inserted into a buffer; see `insert-for-yank' for details.
8571 When a yank handler is specified, STRING must be non-empty (the yank
8572 handler, if non-nil, is stored as a `yank-handler' text property on STRING).
8574 When the yank handler has a non-nil PARAM element, the original STRING
8575 argument is not used by `insert-for-yank'. However, since Lisp code
8576 may access and use elements from the kill ring directly, the STRING
8577 argument should still be a \"useful\" string for such uses."
8580 The @code{kill-new} function looks like this:
8584 (defun kill-new (string &optional replace yank-handler)
8585 "Make STRING the latest kill in the kill ring.
8586 Set `kill-ring-yank-pointer' to point to it.
8588 If `interprogram-cut-function' is non-nil, apply it to STRING.
8589 Optional second argument REPLACE non-nil means that STRING will replace
8590 the front of the kill ring, rather than being added to the list.
8594 (if (> (length string) 0)
8596 (put-text-property 0 (length string)
8597 'yank-handler yank-handler string))
8599 (signal 'args-out-of-range
8600 (list string "yank-handler specified for empty string"))))
8603 (if (fboundp 'menu-bar-update-yank-menu)
8604 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8607 (if (and replace kill-ring)
8608 (setcar kill-ring string)
8609 (push string kill-ring)
8610 (if (> (length kill-ring) kill-ring-max)
8611 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8614 (setq kill-ring-yank-pointer kill-ring)
8615 (if interprogram-cut-function
8616 (funcall interprogram-cut-function string (not replace))))
8621 (defun kill-new (string &optional replace)
8622 "Make STRING the latest kill in the kill ring.
8623 Set the kill-ring-yank pointer to point to it.
8624 If `interprogram-cut-function' is non-nil, apply it to STRING.
8625 Optional second argument REPLACE non-nil means that STRING will replace
8626 the front of the kill ring, rather than being added to the list."
8627 (and (fboundp 'menu-bar-update-yank-menu)
8628 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8629 (if (and replace kill-ring)
8630 (setcar kill-ring string)
8631 (setq kill-ring (cons string kill-ring))
8632 (if (> (length kill-ring) kill-ring-max)
8633 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8634 (setq kill-ring-yank-pointer kill-ring)
8635 (if interprogram-cut-function
8636 (funcall interprogram-cut-function string (not replace))))
8639 (Notice that the function is not interactive.)
8641 As usual, we can look at this function in parts.
8643 The function definition has an optional @code{yank-handler} argument,
8644 which when invoked tells the function how to deal with properties
8645 added to the text, such as `bold' or `italics'. We will skip that.
8648 The first line of the documentation makes sense:
8651 Make STRING the latest kill in the kill ring.
8655 Let's skip over the rest of the documentation for the moment.
8658 Also, let's skip over the initial @code{if} expression and those lines
8659 of code involving @code{menu-bar-update-yank-menu}. We will explain
8663 The critical lines are these:
8667 (if (and replace kill-ring)
8669 (setcar kill-ring string)
8673 (push string kill-ring)
8676 (setq kill-ring (cons string kill-ring))
8677 (if (> (length kill-ring) kill-ring-max)
8678 ;; @r{avoid overly long kill ring}
8679 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8682 (setq kill-ring-yank-pointer kill-ring)
8683 (if interprogram-cut-function
8684 (funcall interprogram-cut-function string (not replace))))
8688 The conditional test is @w{@code{(and replace kill-ring)}}.
8689 This will be true when two conditions are met: the kill ring has
8690 something in it, and the @code{replace} variable is true.
8693 When the @code{kill-append} function sets @code{replace} to be true
8694 and when the kill ring has at least one item in it, the @code{setcar}
8695 expression is executed:
8698 (setcar kill-ring string)
8701 The @code{setcar} function actually changes the first element of the
8702 @code{kill-ring} list to the value of @code{string}. It replaces the
8706 On the other hand, if the kill ring is empty, or replace is false, the
8707 else-part of the condition is executed:
8710 (push string kill-ring)
8715 @code{push} puts its first argument onto the second. It is similar to
8719 (setq kill-ring (cons string kill-ring))
8727 (add-to-list kill-ring string)
8731 When it is false, the expression first constructs a new version of the
8732 kill ring by prepending @code{string} to the existing kill ring as a
8733 new element (that is what the @code{push} does). Then it executes a
8734 second @code{if} clause. This second @code{if} clause keeps the kill
8735 ring from growing too long.
8737 Let's look at these two expressions in order.
8739 The @code{push} line of the else-part sets the new value of the kill
8740 ring to what results from adding the string being killed to the old
8743 We can see how this works with an example.
8749 (setq example-list '("here is a clause" "another clause"))
8754 After evaluating this expression with @kbd{C-x C-e}, you can evaluate
8755 @code{example-list} and see what it returns:
8760 @result{} ("here is a clause" "another clause")
8766 Now, we can add a new element on to this list by evaluating the
8767 following expression:
8768 @findex push, @r{example}
8771 (push "a third clause" example-list)
8776 When we evaluate @code{example-list}, we find its value is:
8781 @result{} ("a third clause" "here is a clause" "another clause")
8786 Thus, the third clause is added to the list by @code{push}.
8789 Now for the second part of the @code{if} clause. This expression
8790 keeps the kill ring from growing too long. It looks like this:
8794 (if (> (length kill-ring) kill-ring-max)
8795 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))
8799 The code checks whether the length of the kill ring is greater than
8800 the maximum permitted length. This is the value of
8801 @code{kill-ring-max} (which is 60, by default). If the length of the
8802 kill ring is too long, then this code sets the last element of the
8803 kill ring to @code{nil}. It does this by using two functions,
8804 @code{nthcdr} and @code{setcdr}.
8806 We looked at @code{setcdr} earlier (@pxref{setcdr, , @code{setcdr}}).
8807 It sets the @sc{cdr} of a list, just as @code{setcar} sets the
8808 @sc{car} of a list. In this case, however, @code{setcdr} will not be
8809 setting the @sc{cdr} of the whole kill ring; the @code{nthcdr}
8810 function is used to cause it to set the @sc{cdr} of the next to last
8811 element of the kill ring---this means that since the @sc{cdr} of the
8812 next to last element is the last element of the kill ring, it will set
8813 the last element of the kill ring.
8815 @findex nthcdr, @r{example}
8816 The @code{nthcdr} function works by repeatedly taking the @sc{cdr} of a
8817 list---it takes the @sc{cdr} of the @sc{cdr} of the @sc{cdr}
8818 @dots{} It does this @var{N} times and returns the results.
8819 (@xref{nthcdr, , @code{nthcdr}}.)
8821 @findex setcdr, @r{example}
8822 Thus, if we had a four element list that was supposed to be three
8823 elements long, we could set the @sc{cdr} of the next to last element
8824 to @code{nil}, and thereby shorten the list. (If you set the last
8825 element to some other value than @code{nil}, which you could do, then
8826 you would not have shortened the list. @xref{setcdr, ,
8829 You can see shortening by evaluating the following three expressions
8830 in turn. First set the value of @code{trees} to @code{(maple oak pine
8831 birch)}, then set the @sc{cdr} of its second @sc{cdr} to @code{nil}
8832 and then find the value of @code{trees}:
8836 (setq trees '(maple oak pine birch))
8837 @result{} (maple oak pine birch)
8841 (setcdr (nthcdr 2 trees) nil)
8845 @result{} (maple oak pine)
8850 (The value returned by the @code{setcdr} expression is @code{nil} since
8851 that is what the @sc{cdr} is set to.)
8853 To repeat, in @code{kill-new}, the @code{nthcdr} function takes the
8854 @sc{cdr} a number of times that is one less than the maximum permitted
8855 size of the kill ring and @code{setcdr} sets the @sc{cdr} of that
8856 element (which will be the rest of the elements in the kill ring) to
8857 @code{nil}. This prevents the kill ring from growing too long.
8860 The next to last expression in the @code{kill-new} function is
8863 (setq kill-ring-yank-pointer kill-ring)
8866 The @code{kill-ring-yank-pointer} is a global variable that is set to be
8867 the @code{kill-ring}.
8869 Even though the @code{kill-ring-yank-pointer} is called a
8870 @samp{pointer}, it is a variable just like the kill ring. However, the
8871 name has been chosen to help humans understand how the variable is used.
8874 Now, to return to an early expression in the body of the function:
8878 (if (fboundp 'menu-bar-update-yank-menu)
8879 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8884 It starts with an @code{if} expression
8886 In this case, the expression tests first to see whether
8887 @code{menu-bar-update-yank-menu} exists as a function, and if so,
8888 calls it. The @code{fboundp} function returns true if the symbol it
8889 is testing has a function definition that `is not void'. If the
8890 symbol's function definition were void, we would receive an error
8891 message, as we did when we created errors intentionally (@pxref{Making
8892 Errors, , Generate an Error Message}).
8895 The then-part contains an expression whose first element is the
8896 function @code{and}.
8899 The @code{and} special form evaluates each of its arguments until one
8900 of the arguments returns a value of @code{nil}, in which case the
8901 @code{and} expression returns @code{nil}; however, if none of the
8902 arguments returns a value of @code{nil}, the value resulting from
8903 evaluating the last argument is returned. (Since such a value is not
8904 @code{nil}, it is considered true in Emacs Lisp.) In other words, an
8905 @code{and} expression returns a true value only if all its arguments
8906 are true. (@xref{Second Buffer Related Review}.)
8908 The expression determines whether the second argument to
8909 @code{menu-bar-update-yank-menu} is true or not.
8911 ;; If we're supposed to be extending an existing string, and that
8912 ;; string really is at the front of the menu, then update it in place.
8915 @code{menu-bar-update-yank-menu} is one of the functions that make it
8916 possible to use the `Select and Paste' menu in the Edit item of a menu
8917 bar; using a mouse, you can look at the various pieces of text you
8918 have saved and select one piece to paste.
8920 The last expression in the @code{kill-new} function adds the newly
8921 copied string to whatever facility exists for copying and pasting
8922 among different programs running in a windowing system. In the X
8923 Windowing system, for example, the @code{x-select-text} function takes
8924 the string and stores it in memory operated by X@. You can paste the
8925 string in another program, such as an Xterm.
8928 The expression looks like this:
8932 (if interprogram-cut-function
8933 (funcall interprogram-cut-function string (not replace))))
8937 If an @code{interprogram-cut-function} exists, then Emacs executes
8938 @code{funcall}, which in turn calls its first argument as a function
8939 and passes the remaining arguments to it. (Incidentally, as far as I
8940 can see, this @code{if} expression could be replaced by an @code{and}
8941 expression similar to the one in the first part of the function.)
8943 We are not going to discuss windowing systems and other programs
8944 further, but merely note that this is a mechanism that enables GNU
8945 Emacs to work easily and well with other programs.
8947 This code for placing text in the kill ring, either concatenated with
8948 an existing element or as a new element, leads us to the code for
8949 bringing back text that has been cut out of the buffer---the yank
8950 commands. However, before discussing the yank commands, it is better
8951 to learn how lists are implemented in a computer. This will make
8952 clear such mysteries as the use of the term `pointer'. But before
8953 that, we will digress into C.
8956 @c is this true in Emacs 22? Does not seems to be
8958 (If the @w{@code{(< end beg))}}
8959 expression is true, @code{kill-append} prepends the string to the just
8960 previously clipped text. For a detailed discussion, see
8961 @ref{kill-append function, , The @code{kill-append} function}.)
8963 If you then yank back the text, i.e., `paste' it, you get both
8964 pieces of text at once. That way, if you delete two words in a row,
8965 and then yank them back, you get both words, in their proper order,
8966 with one yank. (The @w{@code{(< end beg))}} expression makes sure the
8969 On the other hand, if the previous command is not @code{kill-region},
8970 then the @code{kill-new} function is called, which adds the text to
8971 the kill ring as the latest item, and sets the
8972 @code{kill-ring-yank-pointer} variable to point to it.
8976 @c Evidently, changed for Emacs 22. The zap-to-char command does not
8977 @c use the delete-and-extract-region function
8979 2006 Oct 26, the Digression into C is now OK but should come after
8980 copy-region-as-kill and filter-buffer-substring
8984 copy-region-as-kill is short, 12 lines, and uses
8985 filter-buffer-substring, which is longer, 39 lines
8986 and has delete-and-extract-region in it.
8987 delete-and-extract-region is written in C.
8989 see Initializing a Variable with @code{defvar}
8992 @node Digression into C
8993 @section Digression into C
8994 @findex delete-and-extract-region
8995 @cindex C, a digression into
8996 @cindex Digression into C
8998 The @code{copy-region-as-kill} function (@pxref{copy-region-as-kill, ,
8999 @code{copy-region-as-kill}}) uses the @code{filter-buffer-substring}
9000 function, which in turn uses the @code{delete-and-extract-region}
9001 function. It removes the contents of a region and you cannot get them
9004 Unlike the other code discussed here, the
9005 @code{delete-and-extract-region} function is not written in Emacs
9006 Lisp; it is written in C and is one of the primitives of the GNU Emacs
9007 system. Since it is very simple, I will digress briefly from Lisp and
9010 @c GNU Emacs 24 in src/editfns.c
9011 @c the DEFUN for delete-and-extract-region
9014 Like many of the other Emacs primitives,
9015 @code{delete-and-extract-region} is written as an instance of a C
9016 macro, a macro being a template for code. The complete macro looks
9021 DEFUN ("delete-and-extract-region", Fdelete_and_extract_region,
9022 Sdelete_and_extract_region, 2, 2, 0,
9023 doc: /* Delete the text between START and END and return it. */)
9024 (Lisp_Object start, Lisp_Object end)
9026 validate_region (&start, &end);
9027 if (XINT (start) == XINT (end))
9028 return empty_unibyte_string;
9029 return del_range_1 (XINT (start), XINT (end), 1, 1);
9034 Without going into the details of the macro writing process, let me
9035 point out that this macro starts with the word @code{DEFUN}. The word
9036 @code{DEFUN} was chosen since the code serves the same purpose as
9037 @code{defun} does in Lisp. (The @code{DEFUN} C macro is defined in
9038 @file{emacs/src/lisp.h}.)
9040 The word @code{DEFUN} is followed by seven parts inside of
9045 The first part is the name given to the function in Lisp,
9046 @code{delete-and-extract-region}.
9049 The second part is the name of the function in C,
9050 @code{Fdelete_and_extract_region}. By convention, it starts with
9051 @samp{F}. Since C does not use hyphens in names, underscores are used
9055 The third part is the name for the C constant structure that records
9056 information on this function for internal use. It is the name of the
9057 function in C but begins with an @samp{S} instead of an @samp{F}.
9060 The fourth and fifth parts specify the minimum and maximum number of
9061 arguments the function can have. This function demands exactly 2
9065 The sixth part is nearly like the argument that follows the
9066 @code{interactive} declaration in a function written in Lisp: a letter
9067 followed, perhaps, by a prompt. The only difference from the Lisp is
9068 when the macro is called with no arguments. Then you write a @code{0}
9069 (which is a `null string'), as in this macro.
9071 If you were to specify arguments, you would place them between
9072 quotation marks. The C macro for @code{goto-char} includes
9073 @code{"NGoto char: "} in this position to indicate that the function
9074 expects a raw prefix, in this case, a numerical location in a buffer,
9075 and provides a prompt.
9078 The seventh part is a documentation string, just like the one for a
9079 function written in Emacs Lisp. This is written as a C comment. (When
9080 you build Emacs, the program @command{lib-src/make-docfile} extracts
9081 these comments and uses them to make the ``real'' documentation.)
9085 In a C macro, the formal parameters come next, with a statement of
9086 what kind of object they are, followed by what might be called the `body'
9087 of the macro. For @code{delete-and-extract-region} the `body'
9088 consists of the following four lines:
9092 validate_region (&start, &end);
9093 if (XINT (start) == XINT (end))
9094 return empty_unibyte_string;
9095 return del_range_1 (XINT (start), XINT (end), 1, 1);
9099 The @code{validate_region} function checks whether the values
9100 passed as the beginning and end of the region are the proper type and
9101 are within range. If the beginning and end positions are the same,
9102 then return an empty string.
9104 The @code{del_range_1} function actually deletes the text. It is a
9105 complex function we will not look into. It updates the buffer and
9106 does other things. However, it is worth looking at the two arguments
9107 passed to @code{del_range}. These are @w{@code{XINT (start)}} and
9108 @w{@code{XINT (end)}}.
9110 As far as the C language is concerned, @code{start} and @code{end} are
9111 two integers that mark the beginning and end of the region to be
9112 deleted@footnote{More precisely, and requiring more expert knowledge
9113 to understand, the two integers are of type `Lisp_Object', which can
9114 also be a C union instead of an integer type.}.
9116 In early versions of Emacs, these two numbers were thirty-two bits
9117 long, but the code is slowly being generalized to handle other
9118 lengths. Three of the available bits are used to specify the type of
9119 information; the remaining bits are used as `content'.
9121 @samp{XINT} is a C macro that extracts the relevant number from the
9122 longer collection of bits; the three other bits are discarded.
9125 The command in @code{delete-and-extract-region} looks like this:
9128 del_range_1 (XINT (start), XINT (end), 1, 1);
9132 It deletes the region between the beginning position, @code{start},
9133 and the ending position, @code{end}.
9135 From the point of view of the person writing Lisp, Emacs is all very
9136 simple; but hidden underneath is a great deal of complexity to make it
9140 @section Initializing a Variable with @code{defvar}
9142 @cindex Initializing a variable
9143 @cindex Variable initialization
9148 copy-region-as-kill is short, 12 lines, and uses
9149 filter-buffer-substring, which is longer, 39 lines
9150 and has delete-and-extract-region in it.
9151 delete-and-extract-region is written in C.
9153 see Initializing a Variable with @code{defvar}
9157 The @code{copy-region-as-kill} function is written in Emacs Lisp. Two
9158 functions within it, @code{kill-append} and @code{kill-new}, copy a
9159 region in a buffer and save it in a variable called the
9160 @code{kill-ring}. This section describes how the @code{kill-ring}
9161 variable is created and initialized using the @code{defvar} special
9164 (Again we note that the term @code{kill-ring} is a misnomer. The text
9165 that is clipped out of the buffer can be brought back; it is not a ring
9166 of corpses, but a ring of resurrectable text.)
9168 In Emacs Lisp, a variable such as the @code{kill-ring} is created and
9169 given an initial value by using the @code{defvar} special form. The
9170 name comes from ``define variable''.
9172 The @code{defvar} special form is similar to @code{setq} in that it sets
9173 the value of a variable. It is unlike @code{setq} in two ways: first,
9174 it only sets the value of the variable if the variable does not already
9175 have a value. If the variable already has a value, @code{defvar} does
9176 not override the existing value. Second, @code{defvar} has a
9177 documentation string.
9179 (There is a related macro, @code{defcustom}, designed for variables
9180 that people customize. It has more features than @code{defvar}.
9181 (@xref{defcustom, , Setting Variables with @code{defcustom}}.)
9184 * See variable current value::
9185 * defvar and asterisk::
9189 @node See variable current value
9190 @unnumberedsubsec Seeing the Current Value of a Variable
9193 You can see the current value of a variable, any variable, by using
9194 the @code{describe-variable} function, which is usually invoked by
9195 typing @kbd{C-h v}. If you type @kbd{C-h v} and then @code{kill-ring}
9196 (followed by @key{RET}) when prompted, you will see what is in your
9197 current kill ring---this may be quite a lot! Conversely, if you have
9198 been doing nothing this Emacs session except read this document, you
9199 may have nothing in it. Also, you will see the documentation for
9205 List of killed text sequences.
9206 Since the kill ring is supposed to interact nicely with cut-and-paste
9207 facilities offered by window systems, use of this variable should
9210 interact nicely with `interprogram-cut-function' and
9211 `interprogram-paste-function'. The functions `kill-new',
9212 `kill-append', and `current-kill' are supposed to implement this
9213 interaction; you may want to use them instead of manipulating the kill
9219 The kill ring is defined by a @code{defvar} in the following way:
9223 (defvar kill-ring nil
9224 "List of killed text sequences.
9230 In this variable definition, the variable is given an initial value of
9231 @code{nil}, which makes sense, since if you have saved nothing, you want
9232 nothing back if you give a @code{yank} command. The documentation
9233 string is written just like the documentation string of a @code{defun}.
9234 As with the documentation string of the @code{defun}, the first line of
9235 the documentation should be a complete sentence, since some commands,
9236 like @code{apropos}, print only the first line of documentation.
9237 Succeeding lines should not be indented; otherwise they look odd when
9238 you use @kbd{C-h v} (@code{describe-variable}).
9240 @node defvar and asterisk
9241 @subsection @code{defvar} and an asterisk
9242 @findex defvar @r{for a user customizable variable}
9243 @findex defvar @r{with an asterisk}
9245 In the past, Emacs used the @code{defvar} special form both for
9246 internal variables that you would not expect a user to change and for
9247 variables that you do expect a user to change. Although you can still
9248 use @code{defvar} for user customizable variables, please use
9249 @code{defcustom} instead, since it provides a path into
9250 the Customization commands. (@xref{defcustom, , Specifying Variables
9251 using @code{defcustom}}.)
9253 When you specified a variable using the @code{defvar} special form,
9254 you could distinguish a variable that a user might want to change from
9255 others by typing an asterisk, @samp{*}, in the first column of its
9256 documentation string. For example:
9260 (defvar shell-command-default-error-buffer nil
9261 "*Buffer name for `shell-command' @dots{} error output.
9266 @findex set-variable
9268 You could (and still can) use the @code{set-variable} command to
9269 change the value of @code{shell-command-default-error-buffer}
9270 temporarily. However, options set using @code{set-variable} are set
9271 only for the duration of your editing session. The new values are not
9272 saved between sessions. Each time Emacs starts, it reads the original
9273 value, unless you change the value within your @file{.emacs} file,
9274 either by setting it manually or by using @code{customize}.
9275 @xref{Emacs Initialization, , Your @file{.emacs} File}.
9277 For me, the major use of the @code{set-variable} command is to suggest
9278 variables that I might want to set in my @file{.emacs} file. There
9279 are now more than 700 such variables, far too many to remember
9280 readily. Fortunately, you can press @key{TAB} after calling the
9281 @code{M-x set-variable} command to see the list of variables.
9282 (@xref{Examining, , Examining and Setting Variables, emacs,
9283 The GNU Emacs Manual}.)
9286 @node cons & search-fwd Review
9289 Here is a brief summary of some recently introduced functions.
9294 @code{car} returns the first element of a list; @code{cdr} returns the
9295 second and subsequent elements of a list.
9302 (car '(1 2 3 4 5 6 7))
9304 (cdr '(1 2 3 4 5 6 7))
9305 @result{} (2 3 4 5 6 7)
9310 @code{cons} constructs a list by prepending its first argument to its
9324 @code{funcall} evaluates its first argument as a function. It passes
9325 its remaining arguments to its first argument.
9328 Return the result of taking @sc{cdr} `n' times on a list.
9336 The `rest of the rest', as it were.
9343 (nthcdr 3 '(1 2 3 4 5 6 7))
9350 @code{setcar} changes the first element of a list; @code{setcdr}
9351 changes the second and subsequent elements of a list.
9358 (setq triple '(1 2 3))
9365 (setcdr triple '("foo" "bar"))
9368 @result{} (37 "foo" "bar")
9373 Evaluate each argument in sequence and then return the value of the
9386 @item save-restriction
9387 Record whatever narrowing is in effect in the current buffer, if any,
9388 and restore that narrowing after evaluating the arguments.
9390 @item search-forward
9391 Search for a string, and if the string is found, move point. With a
9392 regular expression, use the similar @code{re-search-forward}.
9393 (@xref{Regexp Search, , Regular Expression Searches}, for an
9394 explanation of regular expression patterns and searches.)
9398 @code{search-forward} and @code{re-search-forward} take four
9403 The string or regular expression to search for.
9406 Optionally, the limit of the search.
9409 Optionally, what to do if the search fails, return @code{nil} or an
9413 Optionally, how many times to repeat the search; if negative, the
9414 search goes backwards.
9418 @itemx delete-and-extract-region
9419 @itemx copy-region-as-kill
9421 @code{kill-region} cuts the text between point and mark from the
9422 buffer and stores that text in the kill ring, so you can get it back
9425 @code{copy-region-as-kill} copies the text between point and mark into
9426 the kill ring, from which you can get it by yanking. The function
9427 does not cut or remove the text from the buffer.
9430 @code{delete-and-extract-region} removes the text between point and
9431 mark from the buffer and throws it away. You cannot get it back.
9432 (This is not an interactive command.)
9435 @node search Exercises
9436 @section Searching Exercises
9440 Write an interactive function that searches for a string. If the
9441 search finds the string, leave point after it and display a message
9442 that says ``Found!''. (Do not use @code{search-forward} for the name
9443 of this function; if you do, you will overwrite the existing version of
9444 @code{search-forward} that comes with Emacs. Use a name such as
9445 @code{test-search} instead.)
9448 Write a function that prints the third element of the kill ring in the
9449 echo area, if any; if the kill ring does not contain a third element,
9450 print an appropriate message.
9453 @node List Implementation
9454 @chapter How Lists are Implemented
9455 @cindex Lists in a computer
9457 In Lisp, atoms are recorded in a straightforward fashion; if the
9458 implementation is not straightforward in practice, it is, nonetheless,
9459 straightforward in theory. The atom @samp{rose}, for example, is
9460 recorded as the four contiguous letters @samp{r}, @samp{o}, @samp{s},
9461 @samp{e}. A list, on the other hand, is kept differently. The mechanism
9462 is equally simple, but it takes a moment to get used to the idea. A
9463 list is kept using a series of pairs of pointers. In the series, the
9464 first pointer in each pair points to an atom or to another list, and the
9465 second pointer in each pair points to the next pair, or to the symbol
9466 @code{nil}, which marks the end of the list.
9468 A pointer itself is quite simply the electronic address of what is
9469 pointed to. Hence, a list is kept as a series of electronic addresses.
9472 * Lists diagrammed::
9473 * Symbols as Chest:: Exploring a powerful metaphor.
9478 @node Lists diagrammed
9479 @unnumberedsec Lists diagrammed
9482 For example, the list @code{(rose violet buttercup)} has three elements,
9483 @samp{rose}, @samp{violet}, and @samp{buttercup}. In the computer, the
9484 electronic address of @samp{rose} is recorded in a segment of computer
9485 memory along with the address that gives the electronic address of where
9486 the atom @samp{violet} is located; and that address (the one that tells
9487 where @samp{violet} is located) is kept along with an address that tells
9488 where the address for the atom @samp{buttercup} is located.
9491 This sounds more complicated than it is and is easier seen in a diagram:
9493 @c clear print-postscript-figures
9494 @c !!! cons-cell-diagram #1
9498 ___ ___ ___ ___ ___ ___
9499 |___|___|--> |___|___|--> |___|___|--> nil
9502 --> rose --> violet --> buttercup
9506 @ifset print-postscript-figures
9509 @center @image{cons-1}
9513 @ifclear print-postscript-figures
9517 ___ ___ ___ ___ ___ ___
9518 |___|___|--> |___|___|--> |___|___|--> nil
9521 --> rose --> violet --> buttercup
9528 In the diagram, each box represents a word of computer memory that
9529 holds a Lisp object, usually in the form of a memory address. The boxes,
9530 i.e., the addresses, are in pairs. Each arrow points to what the address
9531 is the address of, either an atom or another pair of addresses. The
9532 first box is the electronic address of @samp{rose} and the arrow points
9533 to @samp{rose}; the second box is the address of the next pair of boxes,
9534 the first part of which is the address of @samp{violet} and the second
9535 part of which is the address of the next pair. The very last box
9536 points to the symbol @code{nil}, which marks the end of the list.
9539 When a variable is set to a list with a function such as @code{setq},
9540 it stores the address of the first box in the variable. Thus,
9541 evaluation of the expression
9544 (setq bouquet '(rose violet buttercup))
9549 creates a situation like this:
9551 @c cons-cell-diagram #2
9557 | ___ ___ ___ ___ ___ ___
9558 --> |___|___|--> |___|___|--> |___|___|--> nil
9561 --> rose --> violet --> buttercup
9565 @ifset print-postscript-figures
9568 @center @image{cons-2}
9572 @ifclear print-postscript-figures
9578 | ___ ___ ___ ___ ___ ___
9579 --> |___|___|--> |___|___|--> |___|___|--> nil
9582 --> rose --> violet --> buttercup
9589 In this example, the symbol @code{bouquet} holds the address of the first
9593 This same list can be illustrated in a different sort of box notation
9596 @c cons-cell-diagram #2a
9602 | -------------- --------------- ----------------
9603 | | car | cdr | | car | cdr | | car | cdr |
9604 -->| rose | o------->| violet | o------->| butter- | nil |
9605 | | | | | | | cup | |
9606 -------------- --------------- ----------------
9610 @ifset print-postscript-figures
9613 @center @image{cons-2a}
9617 @ifclear print-postscript-figures
9623 | -------------- --------------- ----------------
9624 | | car | cdr | | car | cdr | | car | cdr |
9625 -->| rose | o------->| violet | o------->| butter- | nil |
9626 | | | | | | | cup | |
9627 -------------- --------------- ----------------
9633 (Symbols consist of more than pairs of addresses, but the structure of
9634 a symbol is made up of addresses. Indeed, the symbol @code{bouquet}
9635 consists of a group of address-boxes, one of which is the address of
9636 the printed word @samp{bouquet}, a second of which is the address of a
9637 function definition attached to the symbol, if any, a third of which
9638 is the address of the first pair of address-boxes for the list
9639 @code{(rose violet buttercup)}, and so on. Here we are showing that
9640 the symbol's third address-box points to the first pair of
9641 address-boxes for the list.)
9643 If a symbol is set to the @sc{cdr} of a list, the list itself is not
9644 changed; the symbol simply has an address further down the list. (In
9645 the jargon, @sc{car} and @sc{cdr} are `non-destructive'.) Thus,
9646 evaluation of the following expression
9649 (setq flowers (cdr bouquet))
9656 @c cons-cell-diagram #3
9663 | ___ ___ | ___ ___ ___ ___
9664 --> | | | --> | | | | | |
9665 |___|___|----> |___|___|--> |___|___|--> nil
9668 --> rose --> violet --> buttercup
9673 @ifset print-postscript-figures
9676 @center @image{cons-3}
9680 @ifclear print-postscript-figures
9687 | ___ ___ | ___ ___ ___ ___
9688 --> | | | --> | | | | | |
9689 |___|___|----> |___|___|--> |___|___|--> nil
9692 --> rose --> violet --> buttercup
9700 The value of @code{flowers} is @code{(violet buttercup)}, which is
9701 to say, the symbol @code{flowers} holds the address of the pair of
9702 address-boxes, the first of which holds the address of @code{violet},
9703 and the second of which holds the address of @code{buttercup}.
9705 A pair of address-boxes is called a @dfn{cons cell} or @dfn{dotted
9706 pair}. @xref{Cons Cell Type, , Cons Cell and List Types, elisp, The GNU Emacs Lisp
9707 Reference Manual}, and @ref{Dotted Pair Notation, , Dotted Pair
9708 Notation, elisp, The GNU Emacs Lisp Reference Manual}, for more
9709 information about cons cells and dotted pairs.
9712 The function @code{cons} adds a new pair of addresses to the front of
9713 a series of addresses like that shown above. For example, evaluating
9717 (setq bouquet (cons 'lily bouquet))
9724 @c cons-cell-diagram #4
9731 | ___ ___ ___ ___ | ___ ___ ___ ___
9732 --> | | | | | | --> | | | | | |
9733 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9736 --> lily --> rose --> violet --> buttercup
9741 @ifset print-postscript-figures
9744 @center @image{cons-4}
9748 @ifclear print-postscript-figures
9755 | ___ ___ ___ ___ | ___ ___ ___ ___
9756 --> | | | | | | --> | | | | | |
9757 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9760 --> lily --> rose --> violet --> buttercup
9769 However, this does not change the value of the symbol
9770 @code{flowers}, as you can see by evaluating the following,
9773 (eq (cdr (cdr bouquet)) flowers)
9777 which returns @code{t} for true.
9779 Until it is reset, @code{flowers} still has the value
9780 @code{(violet buttercup)}; that is, it has the address of the cons
9781 cell whose first address is of @code{violet}. Also, this does not
9782 alter any of the pre-existing cons cells; they are all still there.
9784 Thus, in Lisp, to get the @sc{cdr} of a list, you just get the address
9785 of the next cons cell in the series; to get the @sc{car} of a list,
9786 you get the address of the first element of the list; to @code{cons} a
9787 new element on a list, you add a new cons cell to the front of the list.
9788 That is all there is to it! The underlying structure of Lisp is
9791 And what does the last address in a series of cons cells refer to? It
9792 is the address of the empty list, of @code{nil}.
9794 In summary, when a Lisp variable is set to a value, it is provided with
9795 the address of the list to which the variable refers.
9797 @node Symbols as Chest
9798 @section Symbols as a Chest of Drawers
9799 @cindex Symbols as a Chest of Drawers
9800 @cindex Chest of Drawers, metaphor for a symbol
9801 @cindex Drawers, Chest of, metaphor for a symbol
9803 In an earlier section, I suggested that you might imagine a symbol as
9804 being a chest of drawers. The function definition is put in one
9805 drawer, the value in another, and so on. What is put in the drawer
9806 holding the value can be changed without affecting the contents of the
9807 drawer holding the function definition, and vice-verse.
9809 Actually, what is put in each drawer is the address of the value or
9810 function definition. It is as if you found an old chest in the attic,
9811 and in one of its drawers you found a map giving you directions to
9812 where the buried treasure lies.
9814 (In addition to its name, symbol definition, and variable value, a
9815 symbol has a `drawer' for a @dfn{property list} which can be used to
9816 record other information. Property lists are not discussed here; see
9817 @ref{Property Lists, , Property Lists, elisp, The GNU Emacs Lisp
9821 Here is a fanciful representation:
9823 @c chest-of-drawers diagram
9828 Chest of Drawers Contents of Drawers
9832 ---------------------
9833 | directions to | [map to]
9834 | symbol name | bouquet
9836 +---------------------+
9838 | symbol definition | [none]
9840 +---------------------+
9841 | directions to | [map to]
9842 | variable value | (rose violet buttercup)
9844 +---------------------+
9846 | property list | [not described here]
9848 +---------------------+
9854 @ifset print-postscript-figures
9857 @center @image{drawers}
9861 @ifclear print-postscript-figures
9866 Chest of Drawers Contents of Drawers
9870 ---------------------
9871 | directions to | [map to]
9872 | symbol name | bouquet
9874 +---------------------+
9876 | symbol definition | [none]
9878 +---------------------+
9879 | directions to | [map to]
9880 | variable value | (rose violet buttercup)
9882 +---------------------+
9884 | property list | [not described here]
9886 +---------------------+
9897 Set @code{flowers} to @code{violet} and @code{buttercup}. Cons two
9898 more flowers on to this list and set this new list to
9899 @code{more-flowers}. Set the @sc{car} of @code{flowers} to a fish.
9900 What does the @code{more-flowers} list now contain?
9903 @chapter Yanking Text Back
9905 @cindex Text retrieval
9906 @cindex Retrieving text
9907 @cindex Pasting text
9909 Whenever you cut text out of a buffer with a `kill' command in GNU Emacs,
9910 you can bring it back with a `yank' command. The text that is cut out of
9911 the buffer is put in the kill ring and the yank commands insert the
9912 appropriate contents of the kill ring back into a buffer (not necessarily
9913 the original buffer).
9915 A simple @kbd{C-y} (@code{yank}) command inserts the first item from
9916 the kill ring into the current buffer. If the @kbd{C-y} command is
9917 followed immediately by @kbd{M-y}, the first element is replaced by
9918 the second element. Successive @kbd{M-y} commands replace the second
9919 element with the third, fourth, or fifth element, and so on. When the
9920 last element in the kill ring is reached, it is replaced by the first
9921 element and the cycle is repeated. (Thus the kill ring is called a
9922 `ring' rather than just a `list'. However, the actual data structure
9923 that holds the text is a list.
9924 @xref{Kill Ring, , Handling the Kill Ring}, for the details of how the
9925 list is handled as a ring.)
9928 * Kill Ring Overview::
9929 * kill-ring-yank-pointer:: The kill ring is a list.
9930 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
9933 @node Kill Ring Overview
9934 @section Kill Ring Overview
9935 @cindex Kill ring overview
9937 The kill ring is a list of textual strings. This is what it looks like:
9940 ("some text" "a different piece of text" "yet more text")
9943 If this were the contents of my kill ring and I pressed @kbd{C-y}, the
9944 string of characters saying @samp{some text} would be inserted in this
9945 buffer where my cursor is located.
9947 The @code{yank} command is also used for duplicating text by copying it.
9948 The copied text is not cut from the buffer, but a copy of it is put on the
9949 kill ring and is inserted by yanking it back.
9951 Three functions are used for bringing text back from the kill ring:
9952 @code{yank}, which is usually bound to @kbd{C-y}; @code{yank-pop},
9953 which is usually bound to @kbd{M-y}; and @code{rotate-yank-pointer},
9954 which is used by the two other functions.
9956 These functions refer to the kill ring through a variable called the
9957 @code{kill-ring-yank-pointer}. Indeed, the insertion code for both the
9958 @code{yank} and @code{yank-pop} functions is:
9961 (insert (car kill-ring-yank-pointer))
9965 (Well, no more. In GNU Emacs 22, the function has been replaced by
9966 @code{insert-for-yank} which calls @code{insert-for-yank-1}
9967 repetitively for each @code{yank-handler} segment. In turn,
9968 @code{insert-for-yank-1} strips text properties from the inserted text
9969 according to @code{yank-excluded-properties}. Otherwise, it is just
9970 like @code{insert}. We will stick with plain @code{insert} since it
9971 is easier to understand.)
9973 To begin to understand how @code{yank} and @code{yank-pop} work, it is
9974 first necessary to look at the @code{kill-ring-yank-pointer} variable.
9976 @node kill-ring-yank-pointer
9977 @section The @code{kill-ring-yank-pointer} Variable
9979 @code{kill-ring-yank-pointer} is a variable, just as @code{kill-ring} is
9980 a variable. It points to something by being bound to the value of what
9981 it points to, like any other Lisp variable.
9984 Thus, if the value of the kill ring is:
9987 ("some text" "a different piece of text" "yet more text")
9992 and the @code{kill-ring-yank-pointer} points to the second clause, the
9993 value of @code{kill-ring-yank-pointer} is:
9996 ("a different piece of text" "yet more text")
9999 As explained in the previous chapter (@pxref{List Implementation}), the
10000 computer does not keep two different copies of the text being pointed to
10001 by both the @code{kill-ring} and the @code{kill-ring-yank-pointer}. The
10002 words ``a different piece of text'' and ``yet more text'' are not
10003 duplicated. Instead, the two Lisp variables point to the same pieces of
10004 text. Here is a diagram:
10006 @c cons-cell-diagram #5
10010 kill-ring kill-ring-yank-pointer
10012 | ___ ___ | ___ ___ ___ ___
10013 ---> | | | --> | | | | | |
10014 |___|___|----> |___|___|--> |___|___|--> nil
10017 | | --> "yet more text"
10019 | --> "a different piece of text"
10026 @ifset print-postscript-figures
10029 @center @image{cons-5}
10033 @ifclear print-postscript-figures
10037 kill-ring kill-ring-yank-pointer
10039 | ___ ___ | ___ ___ ___ ___
10040 ---> | | | --> | | | | | |
10041 |___|___|----> |___|___|--> |___|___|--> nil
10044 | | --> "yet more text"
10046 | --> "a different piece of text
10055 Both the variable @code{kill-ring} and the variable
10056 @code{kill-ring-yank-pointer} are pointers. But the kill ring itself is
10057 usually described as if it were actually what it is composed of. The
10058 @code{kill-ring} is spoken of as if it were the list rather than that it
10059 points to the list. Conversely, the @code{kill-ring-yank-pointer} is
10060 spoken of as pointing to a list.
10062 These two ways of talking about the same thing sound confusing at first but
10063 make sense on reflection. The kill ring is generally thought of as the
10064 complete structure of data that holds the information of what has recently
10065 been cut out of the Emacs buffers. The @code{kill-ring-yank-pointer}
10066 on the other hand, serves to indicate---that is, to `point to'---that part
10067 of the kill ring of which the first element (the @sc{car}) will be
10071 In GNU Emacs 22, the @code{kill-new} function calls
10073 @code{(setq kill-ring-yank-pointer kill-ring)}
10075 (defun rotate-yank-pointer (arg)
10076 "Rotate the yanking point in the kill ring.
10077 With argument, rotate that many kills forward (or backward, if negative)."
10079 (current-kill arg))
10081 (defun current-kill (n &optional do-not-move)
10082 "Rotate the yanking point by N places, and then return that kill.
10083 If N is zero, `interprogram-paste-function' is set, and calling it
10084 returns a string, then that string is added to the front of the
10085 kill ring and returned as the latest kill.
10086 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
10087 yanking point; just return the Nth kill forward."
10088 (let ((interprogram-paste (and (= n 0)
10089 interprogram-paste-function
10090 (funcall interprogram-paste-function))))
10091 (if interprogram-paste
10093 ;; Disable the interprogram cut function when we add the new
10094 ;; text to the kill ring, so Emacs doesn't try to own the
10095 ;; selection, with identical text.
10096 (let ((interprogram-cut-function nil))
10097 (kill-new interprogram-paste))
10098 interprogram-paste)
10099 (or kill-ring (error "Kill ring is empty"))
10100 (let ((ARGth-kill-element
10101 (nthcdr (mod (- n (length kill-ring-yank-pointer))
10102 (length kill-ring))
10105 (setq kill-ring-yank-pointer ARGth-kill-element))
10106 (car ARGth-kill-element)))))
10111 @node yank nthcdr Exercises
10112 @section Exercises with @code{yank} and @code{nthcdr}
10116 Using @kbd{C-h v} (@code{describe-variable}), look at the value of
10117 your kill ring. Add several items to your kill ring; look at its
10118 value again. Using @kbd{M-y} (@code{yank-pop)}, move all the way
10119 around the kill ring. How many items were in your kill ring? Find
10120 the value of @code{kill-ring-max}. Was your kill ring full, or could
10121 you have kept more blocks of text within it?
10124 Using @code{nthcdr} and @code{car}, construct a series of expressions
10125 to return the first, second, third, and fourth elements of a list.
10128 @node Loops & Recursion
10129 @chapter Loops and Recursion
10130 @cindex Loops and recursion
10131 @cindex Recursion and loops
10132 @cindex Repetition (loops)
10134 Emacs Lisp has two primary ways to cause an expression, or a series of
10135 expressions, to be evaluated repeatedly: one uses a @code{while}
10136 loop, and the other uses @dfn{recursion}.
10138 Repetition can be very valuable. For example, to move forward four
10139 sentences, you need only write a program that will move forward one
10140 sentence and then repeat the process four times. Since a computer does
10141 not get bored or tired, such repetitive action does not have the
10142 deleterious effects that excessive or the wrong kinds of repetition can
10145 People mostly write Emacs Lisp functions using @code{while} loops and
10146 their kin; but you can use recursion, which provides a very powerful
10147 way to think about and then to solve problems@footnote{You can write
10148 recursive functions to be frugal or wasteful of mental or computer
10149 resources; as it happens, methods that people find easy---that are
10150 frugal of `mental resources'---sometimes use considerable computer
10151 resources. Emacs was designed to run on machines that we now consider
10152 limited and its default settings are conservative. You may want to
10153 increase the values of @code{max-specpdl-size} and
10154 @code{max-lisp-eval-depth}. In my @file{.emacs} file, I set them to
10155 15 and 30 times their default value.}.
10158 * while:: Causing a stretch of code to repeat.
10160 * Recursion:: Causing a function to call itself.
10161 * Looping exercise::
10165 @section @code{while}
10169 The @code{while} special form tests whether the value returned by
10170 evaluating its first argument is true or false. This is similar to what
10171 the Lisp interpreter does with an @code{if}; what the interpreter does
10172 next, however, is different.
10174 In a @code{while} expression, if the value returned by evaluating the
10175 first argument is false, the Lisp interpreter skips the rest of the
10176 expression (the @dfn{body} of the expression) and does not evaluate it.
10177 However, if the value is true, the Lisp interpreter evaluates the body
10178 of the expression and then again tests whether the first argument to
10179 @code{while} is true or false. If the value returned by evaluating the
10180 first argument is again true, the Lisp interpreter again evaluates the
10181 body of the expression.
10184 The template for a @code{while} expression looks like this:
10188 (while @var{true-or-false-test}
10194 * Looping with while:: Repeat so long as test returns true.
10195 * Loop Example:: A @code{while} loop that uses a list.
10196 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
10197 * Incrementing Loop:: A loop with an incrementing counter.
10198 * Incrementing Loop Details::
10199 * Decrementing Loop:: A loop with a decrementing counter.
10203 @node Looping with while
10204 @unnumberedsubsec Looping with @code{while}
10207 So long as the true-or-false-test of the @code{while} expression
10208 returns a true value when it is evaluated, the body is repeatedly
10209 evaluated. This process is called a loop since the Lisp interpreter
10210 repeats the same thing again and again, like an airplane doing a loop.
10211 When the result of evaluating the true-or-false-test is false, the
10212 Lisp interpreter does not evaluate the rest of the @code{while}
10213 expression and `exits the loop'.
10215 Clearly, if the value returned by evaluating the first argument to
10216 @code{while} is always true, the body following will be evaluated
10217 again and again @dots{} and again @dots{} forever. Conversely, if the
10218 value returned is never true, the expressions in the body will never
10219 be evaluated. The craft of writing a @code{while} loop consists of
10220 choosing a mechanism such that the true-or-false-test returns true
10221 just the number of times that you want the subsequent expressions to
10222 be evaluated, and then have the test return false.
10224 The value returned by evaluating a @code{while} is the value of the
10225 true-or-false-test. An interesting consequence of this is that a
10226 @code{while} loop that evaluates without error will return @code{nil}
10227 or false regardless of whether it has looped 1 or 100 times or none at
10228 all. A @code{while} expression that evaluates successfully never
10229 returns a true value! What this means is that @code{while} is always
10230 evaluated for its side effects, which is to say, the consequences of
10231 evaluating the expressions within the body of the @code{while} loop.
10232 This makes sense. It is not the mere act of looping that is desired,
10233 but the consequences of what happens when the expressions in the loop
10234 are repeatedly evaluated.
10237 @subsection A @code{while} Loop and a List
10239 A common way to control a @code{while} loop is to test whether a list
10240 has any elements. If it does, the loop is repeated; but if it does not,
10241 the repetition is ended. Since this is an important technique, we will
10242 create a short example to illustrate it.
10244 A simple way to test whether a list has elements is to evaluate the
10245 list: if it has no elements, it is an empty list and will return the
10246 empty list, @code{()}, which is a synonym for @code{nil} or false. On
10247 the other hand, a list with elements will return those elements when it
10248 is evaluated. Since Emacs Lisp considers as true any value that is not
10249 @code{nil}, a list that returns elements will test true in a
10253 For example, you can set the variable @code{empty-list} to @code{nil} by
10254 evaluating the following @code{setq} expression:
10257 (setq empty-list ())
10261 After evaluating the @code{setq} expression, you can evaluate the
10262 variable @code{empty-list} in the usual way, by placing the cursor after
10263 the symbol and typing @kbd{C-x C-e}; @code{nil} will appear in your
10270 On the other hand, if you set a variable to be a list with elements, the
10271 list will appear when you evaluate the variable, as you can see by
10272 evaluating the following two expressions:
10276 (setq animals '(gazelle giraffe lion tiger))
10282 Thus, to create a @code{while} loop that tests whether there are any
10283 items in the list @code{animals}, the first part of the loop will be
10294 When the @code{while} tests its first argument, the variable
10295 @code{animals} is evaluated. It returns a list. So long as the list
10296 has elements, the @code{while} considers the results of the test to be
10297 true; but when the list is empty, it considers the results of the test
10300 To prevent the @code{while} loop from running forever, some mechanism
10301 needs to be provided to empty the list eventually. An oft-used
10302 technique is to have one of the subsequent forms in the @code{while}
10303 expression set the value of the list to be the @sc{cdr} of the list.
10304 Each time the @code{cdr} function is evaluated, the list will be made
10305 shorter, until eventually only the empty list will be left. At this
10306 point, the test of the @code{while} loop will return false, and the
10307 arguments to the @code{while} will no longer be evaluated.
10309 For example, the list of animals bound to the variable @code{animals}
10310 can be set to be the @sc{cdr} of the original list with the
10311 following expression:
10314 (setq animals (cdr animals))
10318 If you have evaluated the previous expressions and then evaluate this
10319 expression, you will see @code{(giraffe lion tiger)} appear in the echo
10320 area. If you evaluate the expression again, @code{(lion tiger)} will
10321 appear in the echo area. If you evaluate it again and yet again,
10322 @code{(tiger)} appears and then the empty list, shown by @code{nil}.
10324 A template for a @code{while} loop that uses the @code{cdr} function
10325 repeatedly to cause the true-or-false-test eventually to test false
10330 (while @var{test-whether-list-is-empty}
10332 @var{set-list-to-cdr-of-list})
10336 This test and use of @code{cdr} can be put together in a function that
10337 goes through a list and prints each element of the list on a line of its
10340 @node print-elements-of-list
10341 @subsection An Example: @code{print-elements-of-list}
10342 @findex print-elements-of-list
10344 The @code{print-elements-of-list} function illustrates a @code{while}
10347 @cindex @file{*scratch*} buffer
10348 The function requires several lines for its output. If you are
10349 reading this in a recent instance of GNU Emacs,
10350 @c GNU Emacs 21, GNU Emacs 22, or a later version,
10351 you can evaluate the following expression inside of Info, as usual.
10353 If you are using an earlier version of Emacs, you need to copy the
10354 necessary expressions to your @file{*scratch*} buffer and evaluate
10355 them there. This is because the echo area had only one line in the
10358 You can copy the expressions by marking the beginning of the region
10359 with @kbd{C-@key{SPC}} (@code{set-mark-command}), moving the cursor to
10360 the end of the region and then copying the region using @kbd{M-w}
10361 (@code{kill-ring-save}, which calls @code{copy-region-as-kill} and
10362 then provides visual feedback). In the @file{*scratch*}
10363 buffer, you can yank the expressions back by typing @kbd{C-y}
10366 After you have copied the expressions to the @file{*scratch*} buffer,
10367 evaluate each expression in turn. Be sure to evaluate the last
10368 expression, @code{(print-elements-of-list animals)}, by typing
10369 @kbd{C-u C-x C-e}, that is, by giving an argument to
10370 @code{eval-last-sexp}. This will cause the result of the evaluation
10371 to be printed in the @file{*scratch*} buffer instead of being printed
10372 in the echo area. (Otherwise you will see something like this in your
10373 echo area: @code{^Jgazelle^J^Jgiraffe^J^Jlion^J^Jtiger^Jnil}, in which
10374 each @samp{^J} stands for a `newline'.)
10377 In a recent instance of GNU Emacs, you can evaluate these expressions
10378 directly in the Info buffer, and the echo area will grow to show the
10383 (setq animals '(gazelle giraffe lion tiger))
10385 (defun print-elements-of-list (list)
10386 "Print each element of LIST on a line of its own."
10389 (setq list (cdr list))))
10391 (print-elements-of-list animals)
10397 When you evaluate the three expressions in sequence, you will see
10413 Each element of the list is printed on a line of its own (that is what
10414 the function @code{print} does) and then the value returned by the
10415 function is printed. Since the last expression in the function is the
10416 @code{while} loop, and since @code{while} loops always return
10417 @code{nil}, a @code{nil} is printed after the last element of the list.
10419 @node Incrementing Loop
10420 @subsection A Loop with an Incrementing Counter
10422 A loop is not useful unless it stops when it ought. Besides
10423 controlling a loop with a list, a common way of stopping a loop is to
10424 write the first argument as a test that returns false when the correct
10425 number of repetitions are complete. This means that the loop must
10426 have a counter---an expression that counts how many times the loop
10430 @node Incrementing Loop Details
10431 @unnumberedsubsec Details of an Incrementing Loop
10434 The test for a loop with an incrementing counter can be an expression
10435 such as @code{(< count desired-number)} which returns @code{t} for
10436 true if the value of @code{count} is less than the
10437 @code{desired-number} of repetitions and @code{nil} for false if the
10438 value of @code{count} is equal to or is greater than the
10439 @code{desired-number}. The expression that increments the count can
10440 be a simple @code{setq} such as @code{(setq count (1+ count))}, where
10441 @code{1+} is a built-in function in Emacs Lisp that adds 1 to its
10442 argument. (The expression @w{@code{(1+ count)}} has the same result
10443 as @w{@code{(+ count 1)}}, but is easier for a human to read.)
10446 The template for a @code{while} loop controlled by an incrementing
10447 counter looks like this:
10451 @var{set-count-to-initial-value}
10452 (while (< count desired-number) ; @r{true-or-false-test}
10454 (setq count (1+ count))) ; @r{incrementer}
10459 Note that you need to set the initial value of @code{count}; usually it
10463 * Incrementing Example:: Counting pebbles in a triangle.
10464 * Inc Example parts:: The parts of the function definition.
10465 * Inc Example altogether:: Putting the function definition together.
10468 @node Incrementing Example
10469 @unnumberedsubsubsec Example with incrementing counter
10471 Suppose you are playing on the beach and decide to make a triangle of
10472 pebbles, putting one pebble in the first row, two in the second row,
10473 three in the third row and so on, like this:
10491 @bullet{} @bullet{}
10492 @bullet{} @bullet{} @bullet{}
10493 @bullet{} @bullet{} @bullet{} @bullet{}
10500 (About 2500 years ago, Pythagoras and others developed the beginnings of
10501 number theory by considering questions such as this.)
10503 Suppose you want to know how many pebbles you will need to make a
10504 triangle with 7 rows?
10506 Clearly, what you need to do is add up the numbers from 1 to 7. There
10507 are two ways to do this; start with the smallest number, one, and add up
10508 the list in sequence, 1, 2, 3, 4 and so on; or start with the largest
10509 number and add the list going down: 7, 6, 5, 4 and so on. Because both
10510 mechanisms illustrate common ways of writing @code{while} loops, we will
10511 create two examples, one counting up and the other counting down. In
10512 this first example, we will start with 1 and add 2, 3, 4 and so on.
10514 If you are just adding up a short list of numbers, the easiest way to do
10515 it is to add up all the numbers at once. However, if you do not know
10516 ahead of time how many numbers your list will have, or if you want to be
10517 prepared for a very long list, then you need to design your addition so
10518 that what you do is repeat a simple process many times instead of doing
10519 a more complex process once.
10521 For example, instead of adding up all the pebbles all at once, what you
10522 can do is add the number of pebbles in the first row, 1, to the number
10523 in the second row, 2, and then add the total of those two rows to the
10524 third row, 3. Then you can add the number in the fourth row, 4, to the
10525 total of the first three rows; and so on.
10527 The critical characteristic of the process is that each repetitive
10528 action is simple. In this case, at each step we add only two numbers,
10529 the number of pebbles in the row and the total already found. This
10530 process of adding two numbers is repeated again and again until the last
10531 row has been added to the total of all the preceding rows. In a more
10532 complex loop the repetitive action might not be so simple, but it will
10533 be simpler than doing everything all at once.
10535 @node Inc Example parts
10536 @unnumberedsubsubsec The parts of the function definition
10538 The preceding analysis gives us the bones of our function definition:
10539 first, we will need a variable that we can call @code{total} that will
10540 be the total number of pebbles. This will be the value returned by
10543 Second, we know that the function will require an argument: this
10544 argument will be the total number of rows in the triangle. It can be
10545 called @code{number-of-rows}.
10547 Finally, we need a variable to use as a counter. We could call this
10548 variable @code{counter}, but a better name is @code{row-number}. That
10549 is because what the counter does in this function is count rows, and a
10550 program should be written to be as understandable as possible.
10552 When the Lisp interpreter first starts evaluating the expressions in the
10553 function, the value of @code{total} should be set to zero, since we have
10554 not added anything to it. Then the function should add the number of
10555 pebbles in the first row to the total, and then add the number of
10556 pebbles in the second to the total, and then add the number of
10557 pebbles in the third row to the total, and so on, until there are no
10558 more rows left to add.
10560 Both @code{total} and @code{row-number} are used only inside the
10561 function, so they can be declared as local variables with @code{let}
10562 and given initial values. Clearly, the initial value for @code{total}
10563 should be 0. The initial value of @code{row-number} should be 1,
10564 since we start with the first row. This means that the @code{let}
10565 statement will look like this:
10575 After the internal variables are declared and bound to their initial
10576 values, we can begin the @code{while} loop. The expression that serves
10577 as the test should return a value of @code{t} for true so long as the
10578 @code{row-number} is less than or equal to the @code{number-of-rows}.
10579 (If the expression tests true only so long as the row number is less
10580 than the number of rows in the triangle, the last row will never be
10581 added to the total; hence the row number has to be either less than or
10582 equal to the number of rows.)
10585 @findex <= @r{(less than or equal)}
10586 Lisp provides the @code{<=} function that returns true if the value of
10587 its first argument is less than or equal to the value of its second
10588 argument and false otherwise. So the expression that the @code{while}
10589 will evaluate as its test should look like this:
10592 (<= row-number number-of-rows)
10595 The total number of pebbles can be found by repeatedly adding the number
10596 of pebbles in a row to the total already found. Since the number of
10597 pebbles in the row is equal to the row number, the total can be found by
10598 adding the row number to the total. (Clearly, in a more complex
10599 situation, the number of pebbles in the row might be related to the row
10600 number in a more complicated way; if this were the case, the row number
10601 would be replaced by the appropriate expression.)
10604 (setq total (+ total row-number))
10608 What this does is set the new value of @code{total} to be equal to the
10609 sum of adding the number of pebbles in the row to the previous total.
10611 After setting the value of @code{total}, the conditions need to be
10612 established for the next repetition of the loop, if there is one. This
10613 is done by incrementing the value of the @code{row-number} variable,
10614 which serves as a counter. After the @code{row-number} variable has
10615 been incremented, the true-or-false-test at the beginning of the
10616 @code{while} loop tests whether its value is still less than or equal to
10617 the value of the @code{number-of-rows} and if it is, adds the new value
10618 of the @code{row-number} variable to the @code{total} of the previous
10619 repetition of the loop.
10622 The built-in Emacs Lisp function @code{1+} adds 1 to a number, so the
10623 @code{row-number} variable can be incremented with this expression:
10626 (setq row-number (1+ row-number))
10629 @node Inc Example altogether
10630 @unnumberedsubsubsec Putting the function definition together
10632 We have created the parts for the function definition; now we need to
10636 First, the contents of the @code{while} expression:
10640 (while (<= row-number number-of-rows) ; @r{true-or-false-test}
10641 (setq total (+ total row-number))
10642 (setq row-number (1+ row-number))) ; @r{incrementer}
10646 Along with the @code{let} expression varlist, this very nearly
10647 completes the body of the function definition. However, it requires
10648 one final element, the need for which is somewhat subtle.
10650 The final touch is to place the variable @code{total} on a line by
10651 itself after the @code{while} expression. Otherwise, the value returned
10652 by the whole function is the value of the last expression that is
10653 evaluated in the body of the @code{let}, and this is the value
10654 returned by the @code{while}, which is always @code{nil}.
10656 This may not be evident at first sight. It almost looks as if the
10657 incrementing expression is the last expression of the whole function.
10658 But that expression is part of the body of the @code{while}; it is the
10659 last element of the list that starts with the symbol @code{while}.
10660 Moreover, the whole of the @code{while} loop is a list within the body
10664 In outline, the function will look like this:
10668 (defun @var{name-of-function} (@var{argument-list})
10669 "@var{documentation}@dots{}"
10670 (let (@var{varlist})
10671 (while (@var{true-or-false-test})
10672 @var{body-of-while}@dots{} )
10673 @dots{} )) ; @r{Need final expression here.}
10677 The result of evaluating the @code{let} is what is going to be returned
10678 by the @code{defun} since the @code{let} is not embedded within any
10679 containing list, except for the @code{defun} as a whole. However, if
10680 the @code{while} is the last element of the @code{let} expression, the
10681 function will always return @code{nil}. This is not what we want!
10682 Instead, what we want is the value of the variable @code{total}. This
10683 is returned by simply placing the symbol as the last element of the list
10684 starting with @code{let}. It gets evaluated after the preceding
10685 elements of the list are evaluated, which means it gets evaluated after
10686 it has been assigned the correct value for the total.
10688 It may be easier to see this by printing the list starting with
10689 @code{let} all on one line. This format makes it evident that the
10690 @var{varlist} and @code{while} expressions are the second and third
10691 elements of the list starting with @code{let}, and the @code{total} is
10696 (let (@var{varlist}) (while (@var{true-or-false-test}) @var{body-of-while}@dots{} ) total)
10701 Putting everything together, the @code{triangle} function definition
10706 (defun triangle (number-of-rows) ; @r{Version with}
10707 ; @r{ incrementing counter.}
10708 "Add up the number of pebbles in a triangle.
10709 The first row has one pebble, the second row two pebbles,
10710 the third row three pebbles, and so on.
10711 The argument is NUMBER-OF-ROWS."
10716 (while (<= row-number number-of-rows)
10717 (setq total (+ total row-number))
10718 (setq row-number (1+ row-number)))
10724 After you have installed @code{triangle} by evaluating the function, you
10725 can try it out. Here are two examples:
10736 The sum of the first four numbers is 10 and the sum of the first seven
10739 @node Decrementing Loop
10740 @subsection Loop with a Decrementing Counter
10742 Another common way to write a @code{while} loop is to write the test
10743 so that it determines whether a counter is greater than zero. So long
10744 as the counter is greater than zero, the loop is repeated. But when
10745 the counter is equal to or less than zero, the loop is stopped. For
10746 this to work, the counter has to start out greater than zero and then
10747 be made smaller and smaller by a form that is evaluated
10750 The test will be an expression such as @code{(> counter 0)} which
10751 returns @code{t} for true if the value of @code{counter} is greater
10752 than zero, and @code{nil} for false if the value of @code{counter} is
10753 equal to or less than zero. The expression that makes the number
10754 smaller and smaller can be a simple @code{setq} such as @code{(setq
10755 counter (1- counter))}, where @code{1-} is a built-in function in
10756 Emacs Lisp that subtracts 1 from its argument.
10759 The template for a decrementing @code{while} loop looks like this:
10763 (while (> counter 0) ; @r{true-or-false-test}
10765 (setq counter (1- counter))) ; @r{decrementer}
10770 * Decrementing Example:: More pebbles on the beach.
10771 * Dec Example parts:: The parts of the function definition.
10772 * Dec Example altogether:: Putting the function definition together.
10775 @node Decrementing Example
10776 @unnumberedsubsubsec Example with decrementing counter
10778 To illustrate a loop with a decrementing counter, we will rewrite the
10779 @code{triangle} function so the counter decreases to zero.
10781 This is the reverse of the earlier version of the function. In this
10782 case, to find out how many pebbles are needed to make a triangle with
10783 3 rows, add the number of pebbles in the third row, 3, to the number
10784 in the preceding row, 2, and then add the total of those two rows to
10785 the row that precedes them, which is 1.
10787 Likewise, to find the number of pebbles in a triangle with 7 rows, add
10788 the number of pebbles in the seventh row, 7, to the number in the
10789 preceding row, which is 6, and then add the total of those two rows to
10790 the row that precedes them, which is 5, and so on. As in the previous
10791 example, each addition only involves adding two numbers, the total of
10792 the rows already added up and the number of pebbles in the row that is
10793 being added to the total. This process of adding two numbers is
10794 repeated again and again until there are no more pebbles to add.
10796 We know how many pebbles to start with: the number of pebbles in the
10797 last row is equal to the number of rows. If the triangle has seven
10798 rows, the number of pebbles in the last row is 7. Likewise, we know how
10799 many pebbles are in the preceding row: it is one less than the number in
10802 @node Dec Example parts
10803 @unnumberedsubsubsec The parts of the function definition
10805 We start with three variables: the total number of rows in the
10806 triangle; the number of pebbles in a row; and the total number of
10807 pebbles, which is what we want to calculate. These variables can be
10808 named @code{number-of-rows}, @code{number-of-pebbles-in-row}, and
10809 @code{total}, respectively.
10811 Both @code{total} and @code{number-of-pebbles-in-row} are used only
10812 inside the function and are declared with @code{let}. The initial
10813 value of @code{total} should, of course, be zero. However, the
10814 initial value of @code{number-of-pebbles-in-row} should be equal to
10815 the number of rows in the triangle, since the addition will start with
10819 This means that the beginning of the @code{let} expression will look
10825 (number-of-pebbles-in-row number-of-rows))
10830 The total number of pebbles can be found by repeatedly adding the number
10831 of pebbles in a row to the total already found, that is, by repeatedly
10832 evaluating the following expression:
10835 (setq total (+ total number-of-pebbles-in-row))
10839 After the @code{number-of-pebbles-in-row} is added to the @code{total},
10840 the @code{number-of-pebbles-in-row} should be decremented by one, since
10841 the next time the loop repeats, the preceding row will be
10842 added to the total.
10844 The number of pebbles in a preceding row is one less than the number of
10845 pebbles in a row, so the built-in Emacs Lisp function @code{1-} can be
10846 used to compute the number of pebbles in the preceding row. This can be
10847 done with the following expression:
10851 (setq number-of-pebbles-in-row
10852 (1- number-of-pebbles-in-row))
10856 Finally, we know that the @code{while} loop should stop making repeated
10857 additions when there are no pebbles in a row. So the test for
10858 the @code{while} loop is simply:
10861 (while (> number-of-pebbles-in-row 0)
10864 @node Dec Example altogether
10865 @unnumberedsubsubsec Putting the function definition together
10867 We can put these expressions together to create a function definition
10868 that works. However, on examination, we find that one of the local
10869 variables is unneeded!
10872 The function definition looks like this:
10876 ;;; @r{First subtractive version.}
10877 (defun triangle (number-of-rows)
10878 "Add up the number of pebbles in a triangle."
10880 (number-of-pebbles-in-row number-of-rows))
10881 (while (> number-of-pebbles-in-row 0)
10882 (setq total (+ total number-of-pebbles-in-row))
10883 (setq number-of-pebbles-in-row
10884 (1- number-of-pebbles-in-row)))
10889 As written, this function works.
10891 However, we do not need @code{number-of-pebbles-in-row}.
10893 @cindex Argument as local variable
10894 When the @code{triangle} function is evaluated, the symbol
10895 @code{number-of-rows} will be bound to a number, giving it an initial
10896 value. That number can be changed in the body of the function as if
10897 it were a local variable, without any fear that such a change will
10898 effect the value of the variable outside of the function. This is a
10899 very useful characteristic of Lisp; it means that the variable
10900 @code{number-of-rows} can be used anywhere in the function where
10901 @code{number-of-pebbles-in-row} is used.
10904 Here is a second version of the function written a bit more cleanly:
10908 (defun triangle (number) ; @r{Second version.}
10909 "Return sum of numbers 1 through NUMBER inclusive."
10911 (while (> number 0)
10912 (setq total (+ total number))
10913 (setq number (1- number)))
10918 In brief, a properly written @code{while} loop will consist of three parts:
10922 A test that will return false after the loop has repeated itself the
10923 correct number of times.
10926 An expression the evaluation of which will return the value desired
10927 after being repeatedly evaluated.
10930 An expression to change the value passed to the true-or-false-test so
10931 that the test returns false after the loop has repeated itself the right
10935 @node dolist dotimes
10936 @section Save your time: @code{dolist} and @code{dotimes}
10938 In addition to @code{while}, both @code{dolist} and @code{dotimes}
10939 provide for looping. Sometimes these are quicker to write than the
10940 equivalent @code{while} loop. Both are Lisp macros. (@xref{Macros, ,
10941 Macros, elisp, The GNU Emacs Lisp Reference Manual}. )
10943 @code{dolist} works like a @code{while} loop that `@sc{cdr}s down a
10944 list': @code{dolist} automatically shortens the list each time it
10945 loops---takes the @sc{cdr} of the list---and binds the @sc{car} of
10946 each shorter version of the list to the first of its arguments.
10948 @code{dotimes} loops a specific number of times: you specify the number.
10956 @unnumberedsubsec The @code{dolist} Macro
10959 Suppose, for example, you want to reverse a list, so that
10960 ``first'' ``second'' ``third'' becomes ``third'' ``second'' ``first''.
10963 In practice, you would use the @code{reverse} function, like this:
10967 (setq animals '(gazelle giraffe lion tiger))
10975 Here is how you could reverse the list using a @code{while} loop:
10979 (setq animals '(gazelle giraffe lion tiger))
10981 (defun reverse-list-with-while (list)
10982 "Using while, reverse the order of LIST."
10983 (let (value) ; make sure list starts empty
10985 (setq value (cons (car list) value))
10986 (setq list (cdr list)))
10989 (reverse-list-with-while animals)
10995 And here is how you could use the @code{dolist} macro:
10999 (setq animals '(gazelle giraffe lion tiger))
11001 (defun reverse-list-with-dolist (list)
11002 "Using dolist, reverse the order of LIST."
11003 (let (value) ; make sure list starts empty
11004 (dolist (element list value)
11005 (setq value (cons element value)))))
11007 (reverse-list-with-dolist animals)
11013 In Info, you can place your cursor after the closing parenthesis of
11014 each expression and type @kbd{C-x C-e}; in each case, you should see
11017 (tiger lion giraffe gazelle)
11023 For this example, the existing @code{reverse} function is obviously best.
11024 The @code{while} loop is just like our first example (@pxref{Loop
11025 Example, , A @code{while} Loop and a List}). The @code{while} first
11026 checks whether the list has elements; if so, it constructs a new list
11027 by adding the first element of the list to the existing list (which in
11028 the first iteration of the loop is @code{nil}). Since the second
11029 element is prepended in front of the first element, and the third
11030 element is prepended in front of the second element, the list is reversed.
11032 In the expression using a @code{while} loop,
11033 the @w{@code{(setq list (cdr list))}}
11034 expression shortens the list, so the @code{while} loop eventually
11035 stops. In addition, it provides the @code{cons} expression with a new
11036 first element by creating a new and shorter list at each repetition of
11039 The @code{dolist} expression does very much the same as the
11040 @code{while} expression, except that the @code{dolist} macro does some
11041 of the work you have to do when writing a @code{while} expression.
11043 Like a @code{while} loop, a @code{dolist} loops. What is different is
11044 that it automatically shortens the list each time it loops---it
11045 `@sc{cdr}s down the list' on its own---and it automatically binds
11046 the @sc{car} of each shorter version of the list to the first of its
11049 In the example, the @sc{car} of each shorter version of the list is
11050 referred to using the symbol @samp{element}, the list itself is called
11051 @samp{list}, and the value returned is called @samp{value}. The
11052 remainder of the @code{dolist} expression is the body.
11054 The @code{dolist} expression binds the @sc{car} of each shorter
11055 version of the list to @code{element} and then evaluates the body of
11056 the expression; and repeats the loop. The result is returned in
11060 @unnumberedsubsec The @code{dotimes} Macro
11063 The @code{dotimes} macro is similar to @code{dolist}, except that it
11064 loops a specific number of times.
11066 The first argument to @code{dotimes} is assigned the numbers 0, 1, 2
11067 and so forth each time around the loop, and the value of the third
11068 argument is returned. You need to provide the value of the second
11069 argument, which is how many times the macro loops.
11072 For example, the following binds the numbers from 0 up to, but not
11073 including, the number 3 to the first argument, @var{number}, and then
11074 constructs a list of the three numbers. (The first number is 0, the
11075 second number is 1, and the third number is 2; this makes a total of
11076 three numbers in all, starting with zero as the first number.)
11080 (let (value) ; otherwise a value is a void variable
11081 (dotimes (number 3 value)
11082 (setq value (cons number value))))
11089 @code{dotimes} returns @code{value}, so the way to use
11090 @code{dotimes} is to operate on some expression @var{number} number of
11091 times and then return the result, either as a list or an atom.
11094 Here is an example of a @code{defun} that uses @code{dotimes} to add
11095 up the number of pebbles in a triangle.
11099 (defun triangle-using-dotimes (number-of-rows)
11100 "Using dotimes, add up the number of pebbles in a triangle."
11101 (let ((total 0)) ; otherwise a total is a void variable
11102 (dotimes (number number-of-rows total)
11103 (setq total (+ total (1+ number))))))
11105 (triangle-using-dotimes 4)
11113 A recursive function contains code that tells the Lisp interpreter to
11114 call a program that runs exactly like itself, but with slightly
11115 different arguments. The code runs exactly the same because it has
11116 the same name. However, even though the program has the same name, it
11117 is not the same entity. It is different. In the jargon, it is a
11118 different `instance'.
11120 Eventually, if the program is written correctly, the `slightly
11121 different arguments' will become sufficiently different from the first
11122 arguments that the final instance will stop.
11125 * Building Robots:: Same model, different serial number ...
11126 * Recursive Definition Parts:: Walk until you stop ...
11127 * Recursion with list:: Using a list as the test whether to recurse.
11128 * Recursive triangle function::
11129 * Recursion with cond::
11130 * Recursive Patterns:: Often used templates.
11131 * No Deferment:: Don't store up work ...
11132 * No deferment solution::
11135 @node Building Robots
11136 @subsection Building Robots: Extending the Metaphor
11137 @cindex Building robots
11138 @cindex Robots, building
11140 It is sometimes helpful to think of a running program as a robot that
11141 does a job. In doing its job, a recursive function calls on a second
11142 robot to help it. The second robot is identical to the first in every
11143 way, except that the second robot helps the first and has been
11144 passed different arguments than the first.
11146 In a recursive function, the second robot may call a third; and the
11147 third may call a fourth, and so on. Each of these is a different
11148 entity; but all are clones.
11150 Since each robot has slightly different instructions---the arguments
11151 will differ from one robot to the next---the last robot should know
11154 Let's expand on the metaphor in which a computer program is a robot.
11156 A function definition provides the blueprints for a robot. When you
11157 install a function definition, that is, when you evaluate a
11158 @code{defun} macro, you install the necessary equipment to build
11159 robots. It is as if you were in a factory, setting up an assembly
11160 line. Robots with the same name are built according to the same
11161 blueprints. So they have, as it were, the same `model number', but a
11162 different `serial number'.
11164 We often say that a recursive function `calls itself'. What we mean
11165 is that the instructions in a recursive function cause the Lisp
11166 interpreter to run a different function that has the same name and
11167 does the same job as the first, but with different arguments.
11169 It is important that the arguments differ from one instance to the
11170 next; otherwise, the process will never stop.
11172 @node Recursive Definition Parts
11173 @subsection The Parts of a Recursive Definition
11174 @cindex Parts of a Recursive Definition
11175 @cindex Recursive Definition Parts
11177 A recursive function typically contains a conditional expression which
11182 A true-or-false-test that determines whether the function is called
11183 again, here called the @dfn{do-again-test}.
11186 The name of the function. When this name is called, a new instance of
11187 the function---a new robot, as it were---is created and told what to do.
11190 An expression that returns a different value each time the function is
11191 called, here called the @dfn{next-step-expression}. Consequently, the
11192 argument (or arguments) passed to the new instance of the function
11193 will be different from that passed to the previous instance. This
11194 causes the conditional expression, the @dfn{do-again-test}, to test
11195 false after the correct number of repetitions.
11198 Recursive functions can be much simpler than any other kind of
11199 function. Indeed, when people first start to use them, they often look
11200 so mysteriously simple as to be incomprehensible. Like riding a
11201 bicycle, reading a recursive function definition takes a certain knack
11202 which is hard at first but then seems simple.
11205 There are several different common recursive patterns. A very simple
11206 pattern looks like this:
11210 (defun @var{name-of-recursive-function} (@var{argument-list})
11211 "@var{documentation}@dots{}"
11212 (if @var{do-again-test}
11214 (@var{name-of-recursive-function}
11215 @var{next-step-expression})))
11219 Each time a recursive function is evaluated, a new instance of it is
11220 created and told what to do. The arguments tell the instance what to do.
11222 An argument is bound to the value of the next-step-expression. Each
11223 instance runs with a different value of the next-step-expression.
11225 The value in the next-step-expression is used in the do-again-test.
11227 The value returned by the next-step-expression is passed to the new
11228 instance of the function, which evaluates it (or some
11229 transmogrification of it) to determine whether to continue or stop.
11230 The next-step-expression is designed so that the do-again-test returns
11231 false when the function should no longer be repeated.
11233 The do-again-test is sometimes called the @dfn{stop condition},
11234 since it stops the repetitions when it tests false.
11236 @node Recursion with list
11237 @subsection Recursion with a List
11239 The example of a @code{while} loop that printed the elements of a list
11240 of numbers can be written recursively. Here is the code, including
11241 an expression to set the value of the variable @code{animals} to a list.
11243 If you are reading this in Info in Emacs, you can evaluate this
11244 expression directly in Info. Otherwise, you must copy the example
11245 to the @file{*scratch*} buffer and evaluate each expression there.
11246 Use @kbd{C-u C-x C-e} to evaluate the
11247 @code{(print-elements-recursively animals)} expression so that the
11248 results are printed in the buffer; otherwise the Lisp interpreter will
11249 try to squeeze the results into the one line of the echo area.
11251 Also, place your cursor immediately after the last closing parenthesis
11252 of the @code{print-elements-recursively} function, before the comment.
11253 Otherwise, the Lisp interpreter will try to evaluate the comment.
11255 @findex print-elements-recursively
11258 (setq animals '(gazelle giraffe lion tiger))
11260 (defun print-elements-recursively (list)
11261 "Print each element of LIST on a line of its own.
11263 (when list ; @r{do-again-test}
11264 (print (car list)) ; @r{body}
11265 (print-elements-recursively ; @r{recursive call}
11266 (cdr list)))) ; @r{next-step-expression}
11268 (print-elements-recursively animals)
11272 The @code{print-elements-recursively} function first tests whether
11273 there is any content in the list; if there is, the function prints the
11274 first element of the list, the @sc{car} of the list. Then the
11275 function `invokes itself', but gives itself as its argument, not the
11276 whole list, but the second and subsequent elements of the list, the
11277 @sc{cdr} of the list.
11279 Put another way, if the list is not empty, the function invokes
11280 another instance of code that is similar to the initial code, but is a
11281 different thread of execution, with different arguments than the first
11284 Put in yet another way, if the list is not empty, the first robot
11285 assembles a second robot and tells it what to do; the second robot is
11286 a different individual from the first, but is the same model.
11288 When the second evaluation occurs, the @code{when} expression is
11289 evaluated and if true, prints the first element of the list it
11290 receives as its argument (which is the second element of the original
11291 list). Then the function `calls itself' with the @sc{cdr} of the list
11292 it is invoked with, which (the second time around) is the @sc{cdr} of
11293 the @sc{cdr} of the original list.
11295 Note that although we say that the function `calls itself', what we
11296 mean is that the Lisp interpreter assembles and instructs a new
11297 instance of the program. The new instance is a clone of the first,
11298 but is a separate individual.
11300 Each time the function `invokes itself', it invokes itself on a
11301 shorter version of the original list. It creates a new instance that
11302 works on a shorter list.
11304 Eventually, the function invokes itself on an empty list. It creates
11305 a new instance whose argument is @code{nil}. The conditional expression
11306 tests the value of @code{list}. Since the value of @code{list} is
11307 @code{nil}, the @code{when} expression tests false so the then-part is
11308 not evaluated. The function as a whole then returns @code{nil}.
11311 When you evaluate the expression @code{(print-elements-recursively
11312 animals)} in the @file{*scratch*} buffer, you see this result:
11328 @node Recursive triangle function
11329 @subsection Recursion in Place of a Counter
11330 @findex triangle-recursively
11333 The @code{triangle} function described in a previous section can also
11334 be written recursively. It looks like this:
11338 (defun triangle-recursively (number)
11339 "Return the sum of the numbers 1 through NUMBER inclusive.
11341 (if (= number 1) ; @r{do-again-test}
11343 (+ number ; @r{else-part}
11344 (triangle-recursively ; @r{recursive call}
11345 (1- number))))) ; @r{next-step-expression}
11347 (triangle-recursively 7)
11352 You can install this function by evaluating it and then try it by
11353 evaluating @code{(triangle-recursively 7)}. (Remember to put your
11354 cursor immediately after the last parenthesis of the function
11355 definition, before the comment.) The function evaluates to 28.
11357 To understand how this function works, let's consider what happens in the
11358 various cases when the function is passed 1, 2, 3, or 4 as the value of
11362 * Recursive Example arg of 1 or 2::
11363 * Recursive Example arg of 3 or 4::
11367 @node Recursive Example arg of 1 or 2
11368 @unnumberedsubsubsec An argument of 1 or 2
11371 First, what happens if the value of the argument is 1?
11373 The function has an @code{if} expression after the documentation
11374 string. It tests whether the value of @code{number} is equal to 1; if
11375 so, Emacs evaluates the then-part of the @code{if} expression, which
11376 returns the number 1 as the value of the function. (A triangle with
11377 one row has one pebble in it.)
11379 Suppose, however, that the value of the argument is 2. In this case,
11380 Emacs evaluates the else-part of the @code{if} expression.
11383 The else-part consists of an addition, the recursive call to
11384 @code{triangle-recursively} and a decrementing action; and it looks like
11388 (+ number (triangle-recursively (1- number)))
11391 When Emacs evaluates this expression, the innermost expression is
11392 evaluated first; then the other parts in sequence. Here are the steps
11396 @item Step 1 @w{ } Evaluate the innermost expression.
11398 The innermost expression is @code{(1- number)} so Emacs decrements the
11399 value of @code{number} from 2 to 1.
11401 @item Step 2 @w{ } Evaluate the @code{triangle-recursively} function.
11403 The Lisp interpreter creates an individual instance of
11404 @code{triangle-recursively}. It does not matter that this function is
11405 contained within itself. Emacs passes the result Step 1 as the
11406 argument used by this instance of the @code{triangle-recursively}
11409 In this case, Emacs evaluates @code{triangle-recursively} with an
11410 argument of 1. This means that this evaluation of
11411 @code{triangle-recursively} returns 1.
11413 @item Step 3 @w{ } Evaluate the value of @code{number}.
11415 The variable @code{number} is the second element of the list that
11416 starts with @code{+}; its value is 2.
11418 @item Step 4 @w{ } Evaluate the @code{+} expression.
11420 The @code{+} expression receives two arguments, the first
11421 from the evaluation of @code{number} (Step 3) and the second from the
11422 evaluation of @code{triangle-recursively} (Step 2).
11424 The result of the addition is the sum of 2 plus 1, and the number 3 is
11425 returned, which is correct. A triangle with two rows has three
11429 @node Recursive Example arg of 3 or 4
11430 @unnumberedsubsubsec An argument of 3 or 4
11432 Suppose that @code{triangle-recursively} is called with an argument of
11436 @item Step 1 @w{ } Evaluate the do-again-test.
11438 The @code{if} expression is evaluated first. This is the do-again
11439 test and returns false, so the else-part of the @code{if} expression
11440 is evaluated. (Note that in this example, the do-again-test causes
11441 the function to call itself when it tests false, not when it tests
11444 @item Step 2 @w{ } Evaluate the innermost expression of the else-part.
11446 The innermost expression of the else-part is evaluated, which decrements
11447 3 to 2. This is the next-step-expression.
11449 @item Step 3 @w{ } Evaluate the @code{triangle-recursively} function.
11451 The number 2 is passed to the @code{triangle-recursively} function.
11453 We already know what happens when Emacs evaluates @code{triangle-recursively} with
11454 an argument of 2. After going through the sequence of actions described
11455 earlier, it returns a value of 3. So that is what will happen here.
11457 @item Step 4 @w{ } Evaluate the addition.
11459 3 will be passed as an argument to the addition and will be added to the
11460 number with which the function was called, which is 3.
11464 The value returned by the function as a whole will be 6.
11466 Now that we know what will happen when @code{triangle-recursively} is
11467 called with an argument of 3, it is evident what will happen if it is
11468 called with an argument of 4:
11472 In the recursive call, the evaluation of
11475 (triangle-recursively (1- 4))
11480 will return the value of evaluating
11483 (triangle-recursively 3)
11487 which is 6 and this value will be added to 4 by the addition in the
11492 The value returned by the function as a whole will be 10.
11494 Each time @code{triangle-recursively} is evaluated, it evaluates a
11495 version of itself---a different instance of itself---with a smaller
11496 argument, until the argument is small enough so that it does not
11499 Note that this particular design for a recursive function
11500 requires that operations be deferred.
11502 Before @code{(triangle-recursively 7)} can calculate its answer, it
11503 must call @code{(triangle-recursively 6)}; and before
11504 @code{(triangle-recursively 6)} can calculate its answer, it must call
11505 @code{(triangle-recursively 5)}; and so on. That is to say, the
11506 calculation that @code{(triangle-recursively 7)} makes must be
11507 deferred until @code{(triangle-recursively 6)} makes its calculation;
11508 and @code{(triangle-recursively 6)} must defer until
11509 @code{(triangle-recursively 5)} completes; and so on.
11511 If each of these instances of @code{triangle-recursively} are thought
11512 of as different robots, the first robot must wait for the second to
11513 complete its job, which must wait until the third completes, and so
11516 There is a way around this kind of waiting, which we will discuss in
11517 @ref{No Deferment, , Recursion without Deferments}.
11519 @node Recursion with cond
11520 @subsection Recursion Example Using @code{cond}
11523 The version of @code{triangle-recursively} described earlier is written
11524 with the @code{if} special form. It can also be written using another
11525 special form called @code{cond}. The name of the special form
11526 @code{cond} is an abbreviation of the word @samp{conditional}.
11528 Although the @code{cond} special form is not used as often in the
11529 Emacs Lisp sources as @code{if}, it is used often enough to justify
11533 The template for a @code{cond} expression looks like this:
11543 where the @var{body} is a series of lists.
11546 Written out more fully, the template looks like this:
11551 (@var{first-true-or-false-test} @var{first-consequent})
11552 (@var{second-true-or-false-test} @var{second-consequent})
11553 (@var{third-true-or-false-test} @var{third-consequent})
11558 When the Lisp interpreter evaluates the @code{cond} expression, it
11559 evaluates the first element (the @sc{car} or true-or-false-test) of
11560 the first expression in a series of expressions within the body of the
11563 If the true-or-false-test returns @code{nil} the rest of that
11564 expression, the consequent, is skipped and the true-or-false-test of the
11565 next expression is evaluated. When an expression is found whose
11566 true-or-false-test returns a value that is not @code{nil}, the
11567 consequent of that expression is evaluated. The consequent can be one
11568 or more expressions. If the consequent consists of more than one
11569 expression, the expressions are evaluated in sequence and the value of
11570 the last one is returned. If the expression does not have a consequent,
11571 the value of the true-or-false-test is returned.
11573 If none of the true-or-false-tests test true, the @code{cond} expression
11574 returns @code{nil}.
11577 Written using @code{cond}, the @code{triangle} function looks like this:
11581 (defun triangle-using-cond (number)
11582 (cond ((<= number 0) 0)
11585 (+ number (triangle-using-cond (1- number))))))
11590 In this example, the @code{cond} returns 0 if the number is less than or
11591 equal to 0, it returns 1 if the number is 1 and it evaluates @code{(+
11592 number (triangle-using-cond (1- number)))} if the number is greater than
11595 @node Recursive Patterns
11596 @subsection Recursive Patterns
11597 @cindex Recursive Patterns
11599 Here are three common recursive patterns. Each involves a list.
11600 Recursion does not need to involve lists, but Lisp is designed for lists
11601 and this provides a sense of its primal capabilities.
11610 @unnumberedsubsubsec Recursive Pattern: @emph{every}
11611 @cindex Every, type of recursive pattern
11612 @cindex Recursive pattern: every
11614 In the @code{every} recursive pattern, an action is performed on every
11618 The basic pattern is:
11622 If a list be empty, return @code{nil}.
11624 Else, act on the beginning of the list (the @sc{car} of the list)
11627 through a recursive call by the function on the rest (the
11628 @sc{cdr}) of the list,
11630 and, optionally, combine the acted-on element, using @code{cons},
11631 with the results of acting on the rest.
11640 (defun square-each (numbers-list)
11641 "Square each of a NUMBERS LIST, recursively."
11642 (if (not numbers-list) ; do-again-test
11645 (* (car numbers-list) (car numbers-list))
11646 (square-each (cdr numbers-list))))) ; next-step-expression
11650 (square-each '(1 2 3))
11657 If @code{numbers-list} is empty, do nothing. But if it has content,
11658 construct a list combining the square of the first number in the list
11659 with the result of the recursive call.
11661 (The example follows the pattern exactly: @code{nil} is returned if
11662 the numbers' list is empty. In practice, you would write the
11663 conditional so it carries out the action when the numbers' list is not
11666 The @code{print-elements-recursively} function (@pxref{Recursion with
11667 list, , Recursion with a List}) is another example of an @code{every}
11668 pattern, except in this case, rather than bring the results together
11669 using @code{cons}, we print each element of output.
11672 The @code{print-elements-recursively} function looks like this:
11676 (setq animals '(gazelle giraffe lion tiger))
11680 (defun print-elements-recursively (list)
11681 "Print each element of LIST on a line of its own.
11683 (when list ; @r{do-again-test}
11684 (print (car list)) ; @r{body}
11685 (print-elements-recursively ; @r{recursive call}
11686 (cdr list)))) ; @r{next-step-expression}
11688 (print-elements-recursively animals)
11693 The pattern for @code{print-elements-recursively} is:
11697 When the list is empty, do nothing.
11699 But when the list has at least one element,
11702 act on the beginning of the list (the @sc{car} of the list),
11704 and make a recursive call on the rest (the @sc{cdr}) of the list.
11709 @unnumberedsubsubsec Recursive Pattern: @emph{accumulate}
11710 @cindex Accumulate, type of recursive pattern
11711 @cindex Recursive pattern: accumulate
11713 Another recursive pattern is called the @code{accumulate} pattern. In
11714 the @code{accumulate} recursive pattern, an action is performed on
11715 every element of a list and the result of that action is accumulated
11716 with the results of performing the action on the other elements.
11718 This is very like the `every' pattern using @code{cons}, except that
11719 @code{cons} is not used, but some other combiner.
11726 If a list be empty, return zero or some other constant.
11728 Else, act on the beginning of the list (the @sc{car} of the list),
11731 and combine that acted-on element, using @code{+} or
11732 some other combining function, with
11734 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11739 Here is an example:
11743 (defun add-elements (numbers-list)
11744 "Add the elements of NUMBERS-LIST together."
11745 (if (not numbers-list)
11747 (+ (car numbers-list) (add-elements (cdr numbers-list)))))
11751 (add-elements '(1 2 3 4))
11756 @xref{Files List, , Making a List of Files}, for an example of the
11757 accumulate pattern.
11760 @unnumberedsubsubsec Recursive Pattern: @emph{keep}
11761 @cindex Keep, type of recursive pattern
11762 @cindex Recursive pattern: keep
11764 A third recursive pattern is called the @code{keep} pattern.
11765 In the @code{keep} recursive pattern, each element of a list is tested;
11766 the element is acted on and the results are kept only if the element
11769 Again, this is very like the `every' pattern, except the element is
11770 skipped unless it meets a criterion.
11773 The pattern has three parts:
11777 If a list be empty, return @code{nil}.
11779 Else, if the beginning of the list (the @sc{car} of the list) passes
11783 act on that element and combine it, using @code{cons} with
11785 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11788 Otherwise, if the beginning of the list (the @sc{car} of the list) fails
11792 skip on that element,
11794 and, recursively call the function on the rest (the @sc{cdr}) of the list.
11799 Here is an example that uses @code{cond}:
11803 (defun keep-three-letter-words (word-list)
11804 "Keep three letter words in WORD-LIST."
11806 ;; First do-again-test: stop-condition
11807 ((not word-list) nil)
11809 ;; Second do-again-test: when to act
11810 ((eq 3 (length (symbol-name (car word-list))))
11811 ;; combine acted-on element with recursive call on shorter list
11812 (cons (car word-list) (keep-three-letter-words (cdr word-list))))
11814 ;; Third do-again-test: when to skip element;
11815 ;; recursively call shorter list with next-step expression
11816 (t (keep-three-letter-words (cdr word-list)))))
11820 (keep-three-letter-words '(one two three four five six))
11821 @result{} (one two six)
11825 It goes without saying that you need not use @code{nil} as the test for
11826 when to stop; and you can, of course, combine these patterns.
11829 @subsection Recursion without Deferments
11830 @cindex Deferment in recursion
11831 @cindex Recursion without Deferments
11833 Let's consider again what happens with the @code{triangle-recursively}
11834 function. We will find that the intermediate calculations are
11835 deferred until all can be done.
11838 Here is the function definition:
11842 (defun triangle-recursively (number)
11843 "Return the sum of the numbers 1 through NUMBER inclusive.
11845 (if (= number 1) ; @r{do-again-test}
11847 (+ number ; @r{else-part}
11848 (triangle-recursively ; @r{recursive call}
11849 (1- number))))) ; @r{next-step-expression}
11853 What happens when we call this function with a argument of 7?
11855 The first instance of the @code{triangle-recursively} function adds
11856 the number 7 to the value returned by a second instance of
11857 @code{triangle-recursively}, an instance that has been passed an
11858 argument of 6. That is to say, the first calculation is:
11861 (+ 7 (triangle-recursively 6))
11865 The first instance of @code{triangle-recursively}---you may want to
11866 think of it as a little robot---cannot complete its job. It must hand
11867 off the calculation for @code{(triangle-recursively 6)} to a second
11868 instance of the program, to a second robot. This second individual is
11869 completely different from the first one; it is, in the jargon, a
11870 `different instantiation'. Or, put another way, it is a different
11871 robot. It is the same model as the first; it calculates triangle
11872 numbers recursively; but it has a different serial number.
11874 And what does @code{(triangle-recursively 6)} return? It returns the
11875 number 6 added to the value returned by evaluating
11876 @code{triangle-recursively} with an argument of 5. Using the robot
11877 metaphor, it asks yet another robot to help it.
11883 (+ 7 6 (triangle-recursively 5))
11887 And what happens next?
11890 (+ 7 6 5 (triangle-recursively 4))
11893 Each time @code{triangle-recursively} is called, except for the last
11894 time, it creates another instance of the program---another robot---and
11895 asks it to make a calculation.
11898 Eventually, the full addition is set up and performed:
11904 This design for the function defers the calculation of the first step
11905 until the second can be done, and defers that until the third can be
11906 done, and so on. Each deferment means the computer must remember what
11907 is being waited on. This is not a problem when there are only a few
11908 steps, as in this example. But it can be a problem when there are
11911 @node No deferment solution
11912 @subsection No Deferment Solution
11913 @cindex No deferment solution
11914 @cindex Defermentless solution
11915 @cindex Solution without deferment
11917 The solution to the problem of deferred operations is to write in a
11918 manner that does not defer operations@footnote{The phrase @dfn{tail
11919 recursive} is used to describe such a process, one that uses
11920 `constant space'.}. This requires
11921 writing to a different pattern, often one that involves writing two
11922 function definitions, an `initialization' function and a `helper'
11925 The `initialization' function sets up the job; the `helper' function
11929 Here are the two function definitions for adding up numbers. They are
11930 so simple, I find them hard to understand.
11934 (defun triangle-initialization (number)
11935 "Return the sum of the numbers 1 through NUMBER inclusive.
11936 This is the `initialization' component of a two function
11937 duo that uses recursion."
11938 (triangle-recursive-helper 0 0 number))
11944 (defun triangle-recursive-helper (sum counter number)
11945 "Return SUM, using COUNTER, through NUMBER inclusive.
11946 This is the `helper' component of a two function duo
11947 that uses recursion."
11948 (if (> counter number)
11950 (triangle-recursive-helper (+ sum counter) ; @r{sum}
11951 (1+ counter) ; @r{counter}
11952 number))) ; @r{number}
11957 Install both function definitions by evaluating them, then call
11958 @code{triangle-initialization} with 2 rows:
11962 (triangle-initialization 2)
11967 The `initialization' function calls the first instance of the `helper'
11968 function with three arguments: zero, zero, and a number which is the
11969 number of rows in the triangle.
11971 The first two arguments passed to the `helper' function are
11972 initialization values. These values are changed when
11973 @code{triangle-recursive-helper} invokes new instances.@footnote{The
11974 jargon is mildly confusing: @code{triangle-recursive-helper} uses a
11975 process that is iterative in a procedure that is recursive. The
11976 process is called iterative because the computer need only record the
11977 three values, @code{sum}, @code{counter}, and @code{number}; the
11978 procedure is recursive because the function `calls itself'. On the
11979 other hand, both the process and the procedure used by
11980 @code{triangle-recursively} are called recursive. The word
11981 `recursive' has different meanings in the two contexts.}
11983 Let's see what happens when we have a triangle that has one row. (This
11984 triangle will have one pebble in it!)
11987 @code{triangle-initialization} will call its helper with
11988 the arguments @w{@code{0 0 1}}. That function will run the conditional
11989 test whether @code{(> counter number)}:
11997 and find that the result is false, so it will invoke
11998 the else-part of the @code{if} clause:
12002 (triangle-recursive-helper
12003 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12004 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12005 number) ; @r{number stays the same}
12011 which will first compute:
12015 (triangle-recursive-helper (+ 0 0) ; @r{sum}
12016 (1+ 0) ; @r{counter}
12020 (triangle-recursive-helper 0 1 1)
12024 Again, @code{(> counter number)} will be false, so again, the Lisp
12025 interpreter will evaluate @code{triangle-recursive-helper}, creating a
12026 new instance with new arguments.
12029 This new instance will be;
12033 (triangle-recursive-helper
12034 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12035 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12036 number) ; @r{number stays the same}
12040 (triangle-recursive-helper 1 2 1)
12044 In this case, the @code{(> counter number)} test will be true! So the
12045 instance will return the value of the sum, which will be 1, as
12048 Now, let's pass @code{triangle-initialization} an argument
12049 of 2, to find out how many pebbles there are in a triangle with two rows.
12051 That function calls @code{(triangle-recursive-helper 0 0 2)}.
12054 In stages, the instances called will be:
12058 @r{sum counter number}
12059 (triangle-recursive-helper 0 1 2)
12061 (triangle-recursive-helper 1 2 2)
12063 (triangle-recursive-helper 3 3 2)
12067 When the last instance is called, the @code{(> counter number)} test
12068 will be true, so the instance will return the value of @code{sum},
12071 This kind of pattern helps when you are writing functions that can use
12072 many resources in a computer.
12075 @node Looping exercise
12076 @section Looping Exercise
12080 Write a function similar to @code{triangle} in which each row has a
12081 value which is the square of the row number. Use a @code{while} loop.
12084 Write a function similar to @code{triangle} that multiplies instead of
12088 Rewrite these two functions recursively. Rewrite these functions
12091 @c comma in printed title causes problem in Info cross reference
12093 Write a function for Texinfo mode that creates an index entry at the
12094 beginning of a paragraph for every @samp{@@dfn} within the paragraph.
12095 (In a Texinfo file, @samp{@@dfn} marks a definition. This book is
12096 written in Texinfo.)
12098 Many of the functions you will need are described in two of the
12099 previous chapters, @ref{Cutting & Storing Text, , Cutting and Storing
12100 Text}, and @ref{Yanking, , Yanking Text Back}. If you use
12101 @code{forward-paragraph} to put the index entry at the beginning of
12102 the paragraph, you will have to use @w{@kbd{C-h f}}
12103 (@code{describe-function}) to find out how to make the command go
12106 For more information, see
12108 @ref{Indicating, , Indicating Definitions, texinfo}.
12111 @ref{Indicating, , Indicating, texinfo, Texinfo Manual}, which goes to
12112 a Texinfo manual in the current directory. Or, if you are on the
12114 @uref{http://www.gnu.org/software/texinfo/manual/texinfo/}
12117 ``Indicating Definitions, Commands, etc.'' in @cite{Texinfo, The GNU
12118 Documentation Format}.
12122 @node Regexp Search
12123 @chapter Regular Expression Searches
12124 @cindex Searches, illustrating
12125 @cindex Regular expression searches
12126 @cindex Patterns, searching for
12127 @cindex Motion by sentence and paragraph
12128 @cindex Sentences, movement by
12129 @cindex Paragraphs, movement by
12131 Regular expression searches are used extensively in GNU Emacs. The
12132 two functions, @code{forward-sentence} and @code{forward-paragraph},
12133 illustrate these searches well. They use regular expressions to find
12134 where to move point. The phrase `regular expression' is often written
12137 Regular expression searches are described in @ref{Regexp Search, ,
12138 Regular Expression Search, emacs, The GNU Emacs Manual}, as well as in
12139 @ref{Regular Expressions, , , elisp, The GNU Emacs Lisp Reference
12140 Manual}. In writing this chapter, I am presuming that you have at
12141 least a mild acquaintance with them. The major point to remember is
12142 that regular expressions permit you to search for patterns as well as
12143 for literal strings of characters. For example, the code in
12144 @code{forward-sentence} searches for the pattern of possible
12145 characters that could mark the end of a sentence, and moves point to
12148 Before looking at the code for the @code{forward-sentence} function, it
12149 is worth considering what the pattern that marks the end of a sentence
12150 must be. The pattern is discussed in the next section; following that
12151 is a description of the regular expression search function,
12152 @code{re-search-forward}. The @code{forward-sentence} function
12153 is described in the section following. Finally, the
12154 @code{forward-paragraph} function is described in the last section of
12155 this chapter. @code{forward-paragraph} is a complex function that
12156 introduces several new features.
12159 * sentence-end:: The regular expression for @code{sentence-end}.
12160 * re-search-forward:: Very similar to @code{search-forward}.
12161 * forward-sentence:: A straightforward example of regexp search.
12162 * forward-paragraph:: A somewhat complex example.
12163 * etags:: How to create your own @file{TAGS} table.
12165 * re-search Exercises::
12169 @section The Regular Expression for @code{sentence-end}
12170 @findex sentence-end
12172 The symbol @code{sentence-end} is bound to the pattern that marks the
12173 end of a sentence. What should this regular expression be?
12175 Clearly, a sentence may be ended by a period, a question mark, or an
12176 exclamation mark. Indeed, in English, only clauses that end with one
12177 of those three characters should be considered the end of a sentence.
12178 This means that the pattern should include the character set:
12184 However, we do not want @code{forward-sentence} merely to jump to a
12185 period, a question mark, or an exclamation mark, because such a character
12186 might be used in the middle of a sentence. A period, for example, is
12187 used after abbreviations. So other information is needed.
12189 According to convention, you type two spaces after every sentence, but
12190 only one space after a period, a question mark, or an exclamation mark in
12191 the body of a sentence. So a period, a question mark, or an exclamation
12192 mark followed by two spaces is a good indicator of an end of sentence.
12193 However, in a file, the two spaces may instead be a tab or the end of a
12194 line. This means that the regular expression should include these three
12195 items as alternatives.
12198 This group of alternatives will look like this:
12209 Here, @samp{$} indicates the end of the line, and I have pointed out
12210 where the tab and two spaces are inserted in the expression. Both are
12211 inserted by putting the actual characters into the expression.
12213 Two backslashes, @samp{\\}, are required before the parentheses and
12214 vertical bars: the first backslash quotes the following backslash in
12215 Emacs; and the second indicates that the following character, the
12216 parenthesis or the vertical bar, is special.
12219 Also, a sentence may be followed by one or more carriage returns, like
12230 Like tabs and spaces, a carriage return is inserted into a regular
12231 expression by inserting it literally. The asterisk indicates that the
12232 @key{RET} is repeated zero or more times.
12234 But a sentence end does not consist only of a period, a question mark or
12235 an exclamation mark followed by appropriate space: a closing quotation
12236 mark or a closing brace of some kind may precede the space. Indeed more
12237 than one such mark or brace may precede the space. These require a
12238 expression that looks like this:
12244 In this expression, the first @samp{]} is the first character in the
12245 expression; the second character is @samp{"}, which is preceded by a
12246 @samp{\} to tell Emacs the @samp{"} is @emph{not} special. The last
12247 three characters are @samp{'}, @samp{)}, and @samp{@}}.
12249 All this suggests what the regular expression pattern for matching the
12250 end of a sentence should be; and, indeed, if we evaluate
12251 @code{sentence-end} we find that it returns the following value:
12256 @result{} "[.?!][]\"')@}]*\\($\\| \\| \\)[
12262 (Well, not in GNU Emacs 22; that is because of an effort to make the
12263 process simpler and to handle more glyphs and languages. When the
12264 value of @code{sentence-end} is @code{nil}, then use the value defined
12265 by the function @code{sentence-end}. (Here is a use of the difference
12266 between a value and a function in Emacs Lisp.) The function returns a
12267 value constructed from the variables @code{sentence-end-base},
12268 @code{sentence-end-double-space}, @code{sentence-end-without-period},
12269 and @code{sentence-end-without-space}. The critical variable is
12270 @code{sentence-end-base}; its global value is similar to the one
12271 described above but it also contains two additional quotation marks.
12272 These have differing degrees of curliness. The
12273 @code{sentence-end-without-period} variable, when true, tells Emacs
12274 that a sentence may end without a period, such as text in Thai.)
12278 (Note that here the @key{TAB}, two spaces, and @key{RET} are shown
12279 literally in the pattern.)
12281 This regular expression can be deciphered as follows:
12285 The first part of the pattern is the three characters, a period, a question
12286 mark and an exclamation mark, within square brackets. The pattern must
12287 begin with one or other of these characters.
12290 The second part of the pattern is the group of closing braces and
12291 quotation marks, which can appear zero or more times. These may follow
12292 the period, question mark or exclamation mark. In a regular expression,
12293 the backslash, @samp{\}, followed by the double quotation mark,
12294 @samp{"}, indicates the class of string-quote characters. Usually, the
12295 double quotation mark is the only character in this class. The
12296 asterisk, @samp{*}, indicates that the items in the previous group (the
12297 group surrounded by square brackets, @samp{[]}) may be repeated zero or
12300 @item \\($\\| \\| \\)
12301 The third part of the pattern is one or other of: either the end of a
12302 line, or two blank spaces, or a tab. The double back-slashes are used
12303 to prevent Emacs from reading the parentheses and vertical bars as part
12304 of the search pattern; the parentheses are used to mark the group and
12305 the vertical bars are used to indicated that the patterns to either side
12306 of them are alternatives. The dollar sign is used to indicate the end
12307 of a line and both the two spaces and the tab are each inserted as is to
12308 indicate what they are.
12311 Finally, the last part of the pattern indicates that the end of the line
12312 or the whitespace following the period, question mark or exclamation
12313 mark may, but need not, be followed by one or more carriage returns. In
12314 the pattern, the carriage return is inserted as an actual carriage
12315 return between square brackets but here it is shown as @key{RET}.
12319 @node re-search-forward
12320 @section The @code{re-search-forward} Function
12321 @findex re-search-forward
12323 The @code{re-search-forward} function is very like the
12324 @code{search-forward} function. (@xref{search-forward, , The
12325 @code{search-forward} Function}.)
12327 @code{re-search-forward} searches for a regular expression. If the
12328 search is successful, it leaves point immediately after the last
12329 character in the target. If the search is backwards, it leaves point
12330 just before the first character in the target. You may tell
12331 @code{re-search-forward} to return @code{t} for true. (Moving point
12332 is therefore a `side effect'.)
12334 Like @code{search-forward}, the @code{re-search-forward} function takes
12339 The first argument is the regular expression that the function searches
12340 for. The regular expression will be a string between quotation marks.
12343 The optional second argument limits how far the function will search; it is a
12344 bound, which is specified as a position in the buffer.
12347 The optional third argument specifies how the function responds to
12348 failure: @code{nil} as the third argument causes the function to
12349 signal an error (and print a message) when the search fails; any other
12350 value causes it to return @code{nil} if the search fails and @code{t}
12351 if the search succeeds.
12354 The optional fourth argument is the repeat count. A negative repeat
12355 count causes @code{re-search-forward} to search backwards.
12359 The template for @code{re-search-forward} looks like this:
12363 (re-search-forward "@var{regular-expression}"
12364 @var{limit-of-search}
12365 @var{what-to-do-if-search-fails}
12366 @var{repeat-count})
12370 The second, third, and fourth arguments are optional. However, if you
12371 want to pass a value to either or both of the last two arguments, you
12372 must also pass a value to all the preceding arguments. Otherwise, the
12373 Lisp interpreter will mistake which argument you are passing the value
12377 In the @code{forward-sentence} function, the regular expression will be
12378 the value of the variable @code{sentence-end}. In simple form, that is:
12382 "[.?!][]\"')@}]*\\($\\| \\| \\)[
12388 The limit of the search will be the end of the paragraph (since a
12389 sentence cannot go beyond a paragraph). If the search fails, the
12390 function will return @code{nil}; and the repeat count will be provided
12391 by the argument to the @code{forward-sentence} function.
12393 @node forward-sentence
12394 @section @code{forward-sentence}
12395 @findex forward-sentence
12397 The command to move the cursor forward a sentence is a straightforward
12398 illustration of how to use regular expression searches in Emacs Lisp.
12399 Indeed, the function looks longer and more complicated than it is; this
12400 is because the function is designed to go backwards as well as forwards;
12401 and, optionally, over more than one sentence. The function is usually
12402 bound to the key command @kbd{M-e}.
12405 * Complete forward-sentence::
12406 * fwd-sentence while loops:: Two @code{while} loops.
12407 * fwd-sentence re-search:: A regular expression search.
12411 @node Complete forward-sentence
12412 @unnumberedsubsec Complete @code{forward-sentence} function definition
12416 Here is the code for @code{forward-sentence}:
12421 (defun forward-sentence (&optional arg)
12422 "Move forward to next `sentence-end'. With argument, repeat.
12423 With negative argument, move backward repeatedly to `sentence-beginning'.
12425 The variable `sentence-end' is a regular expression that matches ends of
12426 sentences. Also, every paragraph boundary terminates sentences as well."
12430 (or arg (setq arg 1))
12431 (let ((opoint (point))
12432 (sentence-end (sentence-end)))
12434 (let ((pos (point))
12435 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12436 (if (and (re-search-backward sentence-end par-beg t)
12437 (or (< (match-end 0) pos)
12438 (re-search-backward sentence-end par-beg t)))
12439 (goto-char (match-end 0))
12440 (goto-char par-beg)))
12441 (setq arg (1+ arg)))
12445 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12446 (if (re-search-forward sentence-end par-end t)
12447 (skip-chars-backward " \t\n")
12448 (goto-char par-end)))
12449 (setq arg (1- arg)))
12450 (constrain-to-field nil opoint t)))
12458 (defun forward-sentence (&optional arg)
12459 "Move forward to next sentence-end. With argument, repeat.
12460 With negative argument, move backward repeatedly to sentence-beginning.
12461 Sentence ends are identified by the value of sentence-end
12462 treated as a regular expression. Also, every paragraph boundary
12463 terminates sentences as well."
12467 (or arg (setq arg 1))
12470 (save-excursion (start-of-paragraph-text) (point))))
12471 (if (re-search-backward
12472 (concat sentence-end "[^ \t\n]") par-beg t)
12473 (goto-char (1- (match-end 0)))
12474 (goto-char par-beg)))
12475 (setq arg (1+ arg)))
12478 (save-excursion (end-of-paragraph-text) (point))))
12479 (if (re-search-forward sentence-end par-end t)
12480 (skip-chars-backward " \t\n")
12481 (goto-char par-end)))
12482 (setq arg (1- arg))))
12487 The function looks long at first sight and it is best to look at its
12488 skeleton first, and then its muscle. The way to see the skeleton is to
12489 look at the expressions that start in the left-most columns:
12493 (defun forward-sentence (&optional arg)
12494 "@var{documentation}@dots{}"
12496 (or arg (setq arg 1))
12497 (let ((opoint (point)) (sentence-end (sentence-end)))
12499 (let ((pos (point))
12500 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12501 @var{rest-of-body-of-while-loop-when-going-backwards}
12503 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12504 @var{rest-of-body-of-while-loop-when-going-forwards}
12505 @var{handle-forms-and-equivalent}
12509 This looks much simpler! The function definition consists of
12510 documentation, an @code{interactive} expression, an @code{or}
12511 expression, a @code{let} expression, and @code{while} loops.
12513 Let's look at each of these parts in turn.
12515 We note that the documentation is thorough and understandable.
12517 The function has an @code{interactive "p"} declaration. This means
12518 that the processed prefix argument, if any, is passed to the
12519 function as its argument. (This will be a number.) If the function
12520 is not passed an argument (it is optional) then the argument
12521 @code{arg} will be bound to 1.
12523 When @code{forward-sentence} is called non-interactively without an
12524 argument, @code{arg} is bound to @code{nil}. The @code{or} expression
12525 handles this. What it does is either leave the value of @code{arg} as
12526 it is, but only if @code{arg} is bound to a value; or it sets the
12527 value of @code{arg} to 1, in the case when @code{arg} is bound to
12530 Next is a @code{let}. That specifies the values of two local
12531 variables, @code{point} and @code{sentence-end}. The local value of
12532 point, from before the search, is used in the
12533 @code{constrain-to-field} function which handles forms and
12534 equivalents. The @code{sentence-end} variable is set by the
12535 @code{sentence-end} function.
12537 @node fwd-sentence while loops
12538 @unnumberedsubsec The @code{while} loops
12540 Two @code{while} loops follow. The first @code{while} has a
12541 true-or-false-test that tests true if the prefix argument for
12542 @code{forward-sentence} is a negative number. This is for going
12543 backwards. The body of this loop is similar to the body of the second
12544 @code{while} clause, but it is not exactly the same. We will skip
12545 this @code{while} loop and concentrate on the second @code{while}
12549 The second @code{while} loop is for moving point forward. Its skeleton
12554 (while (> arg 0) ; @r{true-or-false-test}
12556 (if (@var{true-or-false-test})
12559 (setq arg (1- arg)))) ; @code{while} @r{loop decrementer}
12563 The @code{while} loop is of the decrementing kind.
12564 (@xref{Decrementing Loop, , A Loop with a Decrementing Counter}.) It
12565 has a true-or-false-test that tests true so long as the counter (in
12566 this case, the variable @code{arg}) is greater than zero; and it has a
12567 decrementer that subtracts 1 from the value of the counter every time
12570 If no prefix argument is given to @code{forward-sentence}, which is
12571 the most common way the command is used, this @code{while} loop will
12572 run once, since the value of @code{arg} will be 1.
12574 The body of the @code{while} loop consists of a @code{let} expression,
12575 which creates and binds a local variable, and has, as its body, an
12576 @code{if} expression.
12579 The body of the @code{while} loop looks like this:
12584 (save-excursion (end-of-paragraph-text) (point))))
12585 (if (re-search-forward sentence-end par-end t)
12586 (skip-chars-backward " \t\n")
12587 (goto-char par-end)))
12591 The @code{let} expression creates and binds the local variable
12592 @code{par-end}. As we shall see, this local variable is designed to
12593 provide a bound or limit to the regular expression search. If the
12594 search fails to find a proper sentence ending in the paragraph, it will
12595 stop on reaching the end of the paragraph.
12597 But first, let us examine how @code{par-end} is bound to the value of
12598 the end of the paragraph. What happens is that the @code{let} sets the
12599 value of @code{par-end} to the value returned when the Lisp interpreter
12600 evaluates the expression
12604 (save-excursion (end-of-paragraph-text) (point))
12609 In this expression, @code{(end-of-paragraph-text)} moves point to the
12610 end of the paragraph, @code{(point)} returns the value of point, and then
12611 @code{save-excursion} restores point to its original position. Thus,
12612 the @code{let} binds @code{par-end} to the value returned by the
12613 @code{save-excursion} expression, which is the position of the end of
12614 the paragraph. (The @code{end-of-paragraph-text} function uses
12615 @code{forward-paragraph}, which we will discuss shortly.)
12618 Emacs next evaluates the body of the @code{let}, which is an @code{if}
12619 expression that looks like this:
12623 (if (re-search-forward sentence-end par-end t) ; @r{if-part}
12624 (skip-chars-backward " \t\n") ; @r{then-part}
12625 (goto-char par-end))) ; @r{else-part}
12629 The @code{if} tests whether its first argument is true and if so,
12630 evaluates its then-part; otherwise, the Emacs Lisp interpreter
12631 evaluates the else-part. The true-or-false-test of the @code{if}
12632 expression is the regular expression search.
12634 It may seem odd to have what looks like the `real work' of
12635 the @code{forward-sentence} function buried here, but this is a common
12636 way this kind of operation is carried out in Lisp.
12638 @node fwd-sentence re-search
12639 @unnumberedsubsec The regular expression search
12641 The @code{re-search-forward} function searches for the end of the
12642 sentence, that is, for the pattern defined by the @code{sentence-end}
12643 regular expression. If the pattern is found---if the end of the sentence is
12644 found---then the @code{re-search-forward} function does two things:
12648 The @code{re-search-forward} function carries out a side effect, which
12649 is to move point to the end of the occurrence found.
12652 The @code{re-search-forward} function returns a value of true. This is
12653 the value received by the @code{if}, and means that the search was
12658 The side effect, the movement of point, is completed before the
12659 @code{if} function is handed the value returned by the successful
12660 conclusion of the search.
12662 When the @code{if} function receives the value of true from a successful
12663 call to @code{re-search-forward}, the @code{if} evaluates the then-part,
12664 which is the expression @code{(skip-chars-backward " \t\n")}. This
12665 expression moves backwards over any blank spaces, tabs or carriage
12666 returns until a printed character is found and then leaves point after
12667 the character. Since point has already been moved to the end of the
12668 pattern that marks the end of the sentence, this action leaves point
12669 right after the closing printed character of the sentence, which is
12672 On the other hand, if the @code{re-search-forward} function fails to
12673 find a pattern marking the end of the sentence, the function returns
12674 false. The false then causes the @code{if} to evaluate its third
12675 argument, which is @code{(goto-char par-end)}: it moves point to the
12676 end of the paragraph.
12678 (And if the text is in a form or equivalent, and point may not move
12679 fully, then the @code{constrain-to-field} function comes into play.)
12681 Regular expression searches are exceptionally useful and the pattern
12682 illustrated by @code{re-search-forward}, in which the search is the
12683 test of an @code{if} expression, is handy. You will see or write code
12684 incorporating this pattern often.
12686 @node forward-paragraph
12687 @section @code{forward-paragraph}: a Goldmine of Functions
12688 @findex forward-paragraph
12692 (defun forward-paragraph (&optional arg)
12693 "Move forward to end of paragraph.
12694 With argument ARG, do it ARG times;
12695 a negative argument ARG = -N means move backward N paragraphs.
12697 A line which `paragraph-start' matches either separates paragraphs
12698 \(if `paragraph-separate' matches it also) or is the first line of a paragraph.
12699 A paragraph end is the beginning of a line which is not part of the paragraph
12700 to which the end of the previous line belongs, or the end of the buffer.
12701 Returns the count of paragraphs left to move."
12703 (or arg (setq arg 1))
12704 (let* ((opoint (point))
12705 (fill-prefix-regexp
12706 (and fill-prefix (not (equal fill-prefix ""))
12707 (not paragraph-ignore-fill-prefix)
12708 (regexp-quote fill-prefix)))
12709 ;; Remove ^ from paragraph-start and paragraph-sep if they are there.
12710 ;; These regexps shouldn't be anchored, because we look for them
12711 ;; starting at the left-margin. This allows paragraph commands to
12712 ;; work normally with indented text.
12713 ;; This hack will not find problem cases like "whatever\\|^something".
12714 (parstart (if (and (not (equal "" paragraph-start))
12715 (equal ?^ (aref paragraph-start 0)))
12716 (substring paragraph-start 1)
12718 (parsep (if (and (not (equal "" paragraph-separate))
12719 (equal ?^ (aref paragraph-separate 0)))
12720 (substring paragraph-separate 1)
12721 paragraph-separate))
12723 (if fill-prefix-regexp
12724 (concat parsep "\\|"
12725 fill-prefix-regexp "[ \t]*$")
12727 ;; This is used for searching.
12728 (sp-parstart (concat "^[ \t]*\\(?:" parstart "\\|" parsep "\\)"))
12730 (while (and (< arg 0) (not (bobp)))
12731 (if (and (not (looking-at parsep))
12732 (re-search-backward "^\n" (max (1- (point)) (point-min)) t)
12733 (looking-at parsep))
12734 (setq arg (1+ arg))
12735 (setq start (point))
12736 ;; Move back over paragraph-separating lines.
12737 (forward-char -1) (beginning-of-line)
12738 (while (and (not (bobp))
12739 (progn (move-to-left-margin)
12740 (looking-at parsep)))
12744 (setq arg (1+ arg))
12745 ;; Go to end of the previous (non-separating) line.
12747 ;; Search back for line that starts or separates paragraphs.
12748 (if (if fill-prefix-regexp
12749 ;; There is a fill prefix; it overrides parstart.
12750 (let (multiple-lines)
12751 (while (and (progn (beginning-of-line) (not (bobp)))
12752 (progn (move-to-left-margin)
12753 (not (looking-at parsep)))
12754 (looking-at fill-prefix-regexp))
12755 (unless (= (point) start)
12756 (setq multiple-lines t))
12758 (move-to-left-margin)
12759 ;; This deleted code caused a long hanging-indent line
12760 ;; not to be filled together with the following lines.
12761 ;; ;; Don't move back over a line before the paragraph
12762 ;; ;; which doesn't start with fill-prefix
12763 ;; ;; unless that is the only line we've moved over.
12764 ;; (and (not (looking-at fill-prefix-regexp))
12766 ;; (forward-line 1))
12768 (while (and (re-search-backward sp-parstart nil 1)
12769 (setq found-start t)
12770 ;; Found a candidate, but need to check if it is a
12772 (progn (setq start (point))
12773 (move-to-left-margin)
12774 (not (looking-at parsep)))
12775 (not (and (looking-at parstart)
12776 (or (not use-hard-newlines)
12779 (1- start) 'hard)))))
12780 (setq found-start nil)
12785 ;; Move forward over paragraph separators.
12786 ;; We know this cannot reach the place we started
12787 ;; because we know we moved back over a non-separator.
12788 (while (and (not (eobp))
12789 (progn (move-to-left-margin)
12790 (looking-at parsep)))
12792 ;; If line before paragraph is just margin, back up to there.
12794 (if (> (current-column) (current-left-margin))
12796 (skip-chars-backward " \t")
12798 (forward-line 1))))
12799 ;; No starter or separator line => use buffer beg.
12800 (goto-char (point-min))))))
12802 (while (and (> arg 0) (not (eobp)))
12803 ;; Move forward over separator lines...
12804 (while (and (not (eobp))
12805 (progn (move-to-left-margin) (not (eobp)))
12806 (looking-at parsep))
12808 (unless (eobp) (setq arg (1- arg)))
12809 ;; ... and one more line.
12811 (if fill-prefix-regexp
12812 ;; There is a fill prefix; it overrides parstart.
12813 (while (and (not (eobp))
12814 (progn (move-to-left-margin) (not (eobp)))
12815 (not (looking-at parsep))
12816 (looking-at fill-prefix-regexp))
12818 (while (and (re-search-forward sp-parstart nil 1)
12819 (progn (setq start (match-beginning 0))
12822 (progn (move-to-left-margin)
12823 (not (looking-at parsep)))
12824 (or (not (looking-at parstart))
12825 (and use-hard-newlines
12826 (not (get-text-property (1- start) 'hard)))))
12828 (if (< (point) (point-max))
12829 (goto-char start))))
12830 (constrain-to-field nil opoint t)
12831 ;; Return the number of steps that could not be done.
12835 The @code{forward-paragraph} function moves point forward to the end
12836 of the paragraph. It is usually bound to @kbd{M-@}} and makes use of a
12837 number of functions that are important in themselves, including
12838 @code{let*}, @code{match-beginning}, and @code{looking-at}.
12840 The function definition for @code{forward-paragraph} is considerably
12841 longer than the function definition for @code{forward-sentence}
12842 because it works with a paragraph, each line of which may begin with a
12845 A fill prefix consists of a string of characters that are repeated at
12846 the beginning of each line. For example, in Lisp code, it is a
12847 convention to start each line of a paragraph-long comment with
12848 @samp{;;; }. In Text mode, four blank spaces make up another common
12849 fill prefix, creating an indented paragraph. (@xref{Fill Prefix, , ,
12850 emacs, The GNU Emacs Manual}, for more information about fill
12853 The existence of a fill prefix means that in addition to being able to
12854 find the end of a paragraph whose lines begin on the left-most
12855 column, the @code{forward-paragraph} function must be able to find the
12856 end of a paragraph when all or many of the lines in the buffer begin
12857 with the fill prefix.
12859 Moreover, it is sometimes practical to ignore a fill prefix that
12860 exists, especially when blank lines separate paragraphs.
12861 This is an added complication.
12864 * forward-paragraph in brief:: Key parts of the function definition.
12865 * fwd-para let:: The @code{let*} expression.
12866 * fwd-para while:: The forward motion @code{while} loop.
12870 @node forward-paragraph in brief
12871 @unnumberedsubsec Shortened @code{forward-paragraph} function definition
12874 Rather than print all of the @code{forward-paragraph} function, we
12875 will only print parts of it. Read without preparation, the function
12879 In outline, the function looks like this:
12883 (defun forward-paragraph (&optional arg)
12884 "@var{documentation}@dots{}"
12886 (or arg (setq arg 1))
12889 (while (and (< arg 0) (not (bobp))) ; @r{backward-moving-code}
12891 (while (and (> arg 0) (not (eobp))) ; @r{forward-moving-code}
12896 The first parts of the function are routine: the function's argument
12897 list consists of one optional argument. Documentation follows.
12899 The lower case @samp{p} in the @code{interactive} declaration means
12900 that the processed prefix argument, if any, is passed to the function.
12901 This will be a number, and is the repeat count of how many paragraphs
12902 point will move. The @code{or} expression in the next line handles
12903 the common case when no argument is passed to the function, which occurs
12904 if the function is called from other code rather than interactively.
12905 This case was described earlier. (@xref{forward-sentence, The
12906 @code{forward-sentence} function}.) Now we reach the end of the
12907 familiar part of this function.
12910 @unnumberedsubsec The @code{let*} expression
12912 The next line of the @code{forward-paragraph} function begins a
12913 @code{let*} expression. This is a different than @code{let}. The
12914 symbol is @code{let*} not @code{let}.
12916 The @code{let*} special form is like @code{let} except that Emacs sets
12917 each variable in sequence, one after another, and variables in the
12918 latter part of the varlist can make use of the values to which Emacs
12919 set variables in the earlier part of the varlist.
12922 ( refappend save-excursion, , code save-excursion in code append-to-buffer .)
12925 (@ref{append save-excursion, , @code{save-excursion} in @code{append-to-buffer}}.)
12927 In the @code{let*} expression in this function, Emacs binds a total of
12928 seven variables: @code{opoint}, @code{fill-prefix-regexp},
12929 @code{parstart}, @code{parsep}, @code{sp-parstart}, @code{start}, and
12930 @code{found-start}.
12932 The variable @code{parsep} appears twice, first, to remove instances
12933 of @samp{^}, and second, to handle fill prefixes.
12935 The variable @code{opoint} is just the value of @code{point}. As you
12936 can guess, it is used in a @code{constrain-to-field} expression, just
12937 as in @code{forward-sentence}.
12939 The variable @code{fill-prefix-regexp} is set to the value returned by
12940 evaluating the following list:
12945 (not (equal fill-prefix ""))
12946 (not paragraph-ignore-fill-prefix)
12947 (regexp-quote fill-prefix))
12952 This is an expression whose first element is the @code{and} special form.
12954 As we learned earlier (@pxref{kill-new function, , The @code{kill-new}
12955 function}), the @code{and} special form evaluates each of its
12956 arguments until one of the arguments returns a value of @code{nil}, in
12957 which case the @code{and} expression returns @code{nil}; however, if
12958 none of the arguments returns a value of @code{nil}, the value
12959 resulting from evaluating the last argument is returned. (Since such
12960 a value is not @code{nil}, it is considered true in Lisp.) In other
12961 words, an @code{and} expression returns a true value only if all its
12962 arguments are true.
12965 In this case, the variable @code{fill-prefix-regexp} is bound to a
12966 non-@code{nil} value only if the following four expressions produce a
12967 true (i.e., a non-@code{nil}) value when they are evaluated; otherwise,
12968 @code{fill-prefix-regexp} is bound to @code{nil}.
12972 When this variable is evaluated, the value of the fill prefix, if any,
12973 is returned. If there is no fill prefix, this variable returns
12976 @item (not (equal fill-prefix "")
12977 This expression checks whether an existing fill prefix is an empty
12978 string, that is, a string with no characters in it. An empty string is
12979 not a useful fill prefix.
12981 @item (not paragraph-ignore-fill-prefix)
12982 This expression returns @code{nil} if the variable
12983 @code{paragraph-ignore-fill-prefix} has been turned on by being set to a
12984 true value such as @code{t}.
12986 @item (regexp-quote fill-prefix)
12987 This is the last argument to the @code{and} special form. If all the
12988 arguments to the @code{and} are true, the value resulting from
12989 evaluating this expression will be returned by the @code{and} expression
12990 and bound to the variable @code{fill-prefix-regexp},
12993 @findex regexp-quote
12995 The result of evaluating this @code{and} expression successfully is that
12996 @code{fill-prefix-regexp} will be bound to the value of
12997 @code{fill-prefix} as modified by the @code{regexp-quote} function.
12998 What @code{regexp-quote} does is read a string and return a regular
12999 expression that will exactly match the string and match nothing else.
13000 This means that @code{fill-prefix-regexp} will be set to a value that
13001 will exactly match the fill prefix if the fill prefix exists.
13002 Otherwise, the variable will be set to @code{nil}.
13004 The next two local variables in the @code{let*} expression are
13005 designed to remove instances of @samp{^} from @code{parstart} and
13006 @code{parsep}, the local variables which indicate the paragraph start
13007 and the paragraph separator. The next expression sets @code{parsep}
13008 again. That is to handle fill prefixes.
13010 This is the setting that requires the definition call @code{let*}
13011 rather than @code{let}. The true-or-false-test for the @code{if}
13012 depends on whether the variable @code{fill-prefix-regexp} evaluates to
13013 @code{nil} or some other value.
13015 If @code{fill-prefix-regexp} does not have a value, Emacs evaluates
13016 the else-part of the @code{if} expression and binds @code{parsep} to
13017 its local value. (@code{parsep} is a regular expression that matches
13018 what separates paragraphs.)
13020 But if @code{fill-prefix-regexp} does have a value, Emacs evaluates
13021 the then-part of the @code{if} expression and binds @code{parsep} to a
13022 regular expression that includes the @code{fill-prefix-regexp} as part
13025 Specifically, @code{parsep} is set to the original value of the
13026 paragraph separate regular expression concatenated with an alternative
13027 expression that consists of the @code{fill-prefix-regexp} followed by
13028 optional whitespace to the end of the line. The whitespace is defined
13029 by @w{@code{"[ \t]*$"}}.) The @samp{\\|} defines this portion of the
13030 regexp as an alternative to @code{parsep}.
13032 According to a comment in the code, the next local variable,
13033 @code{sp-parstart}, is used for searching, and then the final two,
13034 @code{start} and @code{found-start}, are set to @code{nil}.
13036 Now we get into the body of the @code{let*}. The first part of the body
13037 of the @code{let*} deals with the case when the function is given a
13038 negative argument and is therefore moving backwards. We will skip this
13041 @node fwd-para while
13042 @unnumberedsubsec The forward motion @code{while} loop
13044 The second part of the body of the @code{let*} deals with forward
13045 motion. It is a @code{while} loop that repeats itself so long as the
13046 value of @code{arg} is greater than zero. In the most common use of
13047 the function, the value of the argument is 1, so the body of the
13048 @code{while} loop is evaluated exactly once, and the cursor moves
13049 forward one paragraph.
13052 (while (and (> arg 0) (not (eobp)))
13054 ;; Move forward over separator lines...
13055 (while (and (not (eobp))
13056 (progn (move-to-left-margin) (not (eobp)))
13057 (looking-at parsep))
13059 (unless (eobp) (setq arg (1- arg)))
13060 ;; ... and one more line.
13063 (if fill-prefix-regexp
13064 ;; There is a fill prefix; it overrides parstart.
13065 (while (and (not (eobp))
13066 (progn (move-to-left-margin) (not (eobp)))
13067 (not (looking-at parsep))
13068 (looking-at fill-prefix-regexp))
13071 (while (and (re-search-forward sp-parstart nil 1)
13072 (progn (setq start (match-beginning 0))
13075 (progn (move-to-left-margin)
13076 (not (looking-at parsep)))
13077 (or (not (looking-at parstart))
13078 (and use-hard-newlines
13079 (not (get-text-property (1- start) 'hard)))))
13082 (if (< (point) (point-max))
13083 (goto-char start))))
13086 This part handles three situations: when point is between paragraphs,
13087 when there is a fill prefix and when there is no fill prefix.
13090 The @code{while} loop looks like this:
13094 ;; @r{going forwards and not at the end of the buffer}
13095 (while (and (> arg 0) (not (eobp)))
13097 ;; @r{between paragraphs}
13098 ;; Move forward over separator lines...
13099 (while (and (not (eobp))
13100 (progn (move-to-left-margin) (not (eobp)))
13101 (looking-at parsep))
13103 ;; @r{This decrements the loop}
13104 (unless (eobp) (setq arg (1- arg)))
13105 ;; ... and one more line.
13110 (if fill-prefix-regexp
13111 ;; There is a fill prefix; it overrides parstart;
13112 ;; we go forward line by line
13113 (while (and (not (eobp))
13114 (progn (move-to-left-margin) (not (eobp)))
13115 (not (looking-at parsep))
13116 (looking-at fill-prefix-regexp))
13121 ;; There is no fill prefix;
13122 ;; we go forward character by character
13123 (while (and (re-search-forward sp-parstart nil 1)
13124 (progn (setq start (match-beginning 0))
13127 (progn (move-to-left-margin)
13128 (not (looking-at parsep)))
13129 (or (not (looking-at parstart))
13130 (and use-hard-newlines
13131 (not (get-text-property (1- start) 'hard)))))
13136 ;; and if there is no fill prefix and if we are not at the end,
13137 ;; go to whatever was found in the regular expression search
13139 (if (< (point) (point-max))
13140 (goto-char start))))
13145 We can see that this is a decrementing counter @code{while} loop,
13146 using the expression @code{(setq arg (1- arg))} as the decrementer.
13147 That expression is not far from the @code{while}, but is hidden in
13148 another Lisp macro, an @code{unless} macro. Unless we are at the end
13149 of the buffer---that is what the @code{eobp} function determines; it
13150 is an abbreviation of @samp{End Of Buffer P}---we decrease the value
13151 of @code{arg} by one.
13153 (If we are at the end of the buffer, we cannot go forward any more and
13154 the next loop of the @code{while} expression will test false since the
13155 test is an @code{and} with @code{(not (eobp))}. The @code{not}
13156 function means exactly as you expect; it is another name for
13157 @code{null}, a function that returns true when its argument is false.)
13159 Interestingly, the loop count is not decremented until we leave the
13160 space between paragraphs, unless we come to the end of buffer or stop
13161 seeing the local value of the paragraph separator.
13163 That second @code{while} also has a @code{(move-to-left-margin)}
13164 expression. The function is self-explanatory. It is inside a
13165 @code{progn} expression and not the last element of its body, so it is
13166 only invoked for its side effect, which is to move point to the left
13167 margin of the current line.
13170 The @code{looking-at} function is also self-explanatory; it returns
13171 true if the text after point matches the regular expression given as
13174 The rest of the body of the loop looks difficult at first, but makes
13175 sense as you come to understand it.
13178 First consider what happens if there is a fill prefix:
13182 (if fill-prefix-regexp
13183 ;; There is a fill prefix; it overrides parstart;
13184 ;; we go forward line by line
13185 (while (and (not (eobp))
13186 (progn (move-to-left-margin) (not (eobp)))
13187 (not (looking-at parsep))
13188 (looking-at fill-prefix-regexp))
13194 This expression moves point forward line by line so long
13195 as four conditions are true:
13199 Point is not at the end of the buffer.
13202 We can move to the left margin of the text and are
13203 not at the end of the buffer.
13206 The text following point does not separate paragraphs.
13209 The pattern following point is the fill prefix regular expression.
13212 The last condition may be puzzling, until you remember that point was
13213 moved to the beginning of the line early in the @code{forward-paragraph}
13214 function. This means that if the text has a fill prefix, the
13215 @code{looking-at} function will see it.
13218 Consider what happens when there is no fill prefix.
13222 (while (and (re-search-forward sp-parstart nil 1)
13223 (progn (setq start (match-beginning 0))
13226 (progn (move-to-left-margin)
13227 (not (looking-at parsep)))
13228 (or (not (looking-at parstart))
13229 (and use-hard-newlines
13230 (not (get-text-property (1- start) 'hard)))))
13236 This @code{while} loop has us searching forward for
13237 @code{sp-parstart}, which is the combination of possible whitespace
13238 with a the local value of the start of a paragraph or of a paragraph
13239 separator. (The latter two are within an expression starting
13240 @code{\(?:} so that they are not referenced by the
13241 @code{match-beginning} function.)
13244 The two expressions,
13248 (setq start (match-beginning 0))
13254 mean go to the start of the text matched by the regular expression
13257 The @code{(match-beginning 0)} expression is new. It returns a number
13258 specifying the location of the start of the text that was matched by
13261 The @code{match-beginning} function is used here because of a
13262 characteristic of a forward search: a successful forward search,
13263 regardless of whether it is a plain search or a regular expression
13264 search, moves point to the end of the text that is found. In this
13265 case, a successful search moves point to the end of the pattern for
13266 @code{sp-parstart}.
13268 However, we want to put point at the end of the current paragraph, not
13269 somewhere else. Indeed, since the search possibly includes the
13270 paragraph separator, point may end up at the beginning of the next one
13271 unless we use an expression that includes @code{match-beginning}.
13273 @findex match-beginning
13274 When given an argument of 0, @code{match-beginning} returns the
13275 position that is the start of the text matched by the most recent
13276 search. In this case, the most recent search looks for
13277 @code{sp-parstart}. The @code{(match-beginning 0)} expression returns
13278 the beginning position of that pattern, rather than the end position
13281 (Incidentally, when passed a positive number as an argument, the
13282 @code{match-beginning} function returns the location of point at that
13283 parenthesized expression in the last search unless that parenthesized
13284 expression begins with @code{\(?:}. I don't know why @code{\(?:}
13285 appears here since the argument is 0.)
13288 The last expression when there is no fill prefix is
13292 (if (< (point) (point-max))
13293 (goto-char start))))
13298 This says that if there is no fill prefix and if we are not at the
13299 end, point should move to the beginning of whatever was found by the
13300 regular expression search for @code{sp-parstart}.
13302 The full definition for the @code{forward-paragraph} function not only
13303 includes code for going forwards, but also code for going backwards.
13305 If you are reading this inside of GNU Emacs and you want to see the
13306 whole function, you can type @kbd{C-h f} (@code{describe-function})
13307 and the name of the function. This gives you the function
13308 documentation and the name of the library containing the function's
13309 source. Place point over the name of the library and press the RET
13310 key; you will be taken directly to the source. (Be sure to install
13311 your sources! Without them, you are like a person who tries to drive
13312 a car with his eyes shut!)
13315 @section Create Your Own @file{TAGS} File
13317 @cindex @file{TAGS} file, create own
13319 Besides @kbd{C-h f} (@code{describe-function}), another way to see the
13320 source of a function is to type @kbd{M-.} (@code{find-tag}) and the
13321 name of the function when prompted for it. This is a good habit to
13322 get into. The @kbd{M-.} (@code{find-tag}) command takes you directly
13323 to the source for a function, variable, or node. The function depends
13324 on tags tables to tell it where to go.
13326 If the @code{find-tag} function first asks you for the name of a
13327 @file{TAGS} table, give it the name of a @file{TAGS} file such as
13328 @file{/usr/local/src/emacs/src/TAGS}. (The exact path to your
13329 @file{TAGS} file depends on how your copy of Emacs was installed. I
13330 just told you the location that provides both my C and my Emacs Lisp
13333 You can also create your own @file{TAGS} file for directories that
13336 You often need to build and install tags tables yourself. They are
13337 not built automatically. A tags table is called a @file{TAGS} file;
13338 the name is in upper case letters.
13340 You can create a @file{TAGS} file by calling the @code{etags} program
13341 that comes as a part of the Emacs distribution. Usually, @code{etags}
13342 is compiled and installed when Emacs is built. (@code{etags} is not
13343 an Emacs Lisp function or a part of Emacs; it is a C program.)
13346 To create a @file{TAGS} file, first switch to the directory in which
13347 you want to create the file. In Emacs you can do this with the
13348 @kbd{M-x cd} command, or by visiting a file in the directory, or by
13349 listing the directory with @kbd{C-x d} (@code{dired}). Then run the
13350 compile command, with @w{@code{etags *.el}} as the command to execute
13353 M-x compile RET etags *.el RET
13357 to create a @file{TAGS} file for Emacs Lisp.
13359 For example, if you have a large number of files in your
13360 @file{~/emacs} directory, as I do---I have 137 @file{.el} files in it,
13361 of which I load 12---you can create a @file{TAGS} file for the Emacs
13362 Lisp files in that directory.
13365 The @code{etags} program takes all the usual shell `wildcards'. For
13366 example, if you have two directories for which you want a single
13367 @file{TAGS} file, type @w{@code{etags *.el ../elisp/*.el}}, where
13368 @file{../elisp/} is the second directory:
13371 M-x compile RET etags *.el ../elisp/*.el RET
13378 M-x compile RET etags --help RET
13382 to see a list of the options accepted by @code{etags} as well as a
13383 list of supported languages.
13385 The @code{etags} program handles more than 20 languages, including
13386 Emacs Lisp, Common Lisp, Scheme, C, C++, Ada, Fortran, HTML, Java,
13387 LaTeX, Pascal, Perl, PostScript, Python, TeX, Texinfo, makefiles, and
13388 most assemblers. The program has no switches for specifying the
13389 language; it recognizes the language in an input file according to its
13390 file name and contents.
13392 @file{etags} is very helpful when you are writing code yourself and
13393 want to refer back to functions you have already written. Just run
13394 @code{etags} again at intervals as you write new functions, so they
13395 become part of the @file{TAGS} file.
13397 If you think an appropriate @file{TAGS} file already exists for what
13398 you want, but do not know where it is, you can use the @code{locate}
13399 program to attempt to find it.
13401 Type @w{@kbd{M-x locate @key{RET} TAGS @key{RET}}} and Emacs will list
13402 for you the full path names of all your @file{TAGS} files. On my
13403 system, this command lists 34 @file{TAGS} files. On the other hand, a
13404 `plain vanilla' system I recently installed did not contain any
13407 If the tags table you want has been created, you can use the @code{M-x
13408 visit-tags-table} command to specify it. Otherwise, you will need to
13409 create the tag table yourself and then use @code{M-x
13412 @subsubheading Building Tags in the Emacs sources
13413 @cindex Building Tags in the Emacs sources
13414 @cindex Tags in the Emacs sources
13417 The GNU Emacs sources come with a @file{Makefile} that contains a
13418 sophisticated @code{etags} command that creates, collects, and merges
13419 tags tables from all over the Emacs sources and puts the information
13420 into one @file{TAGS} file in the @file{src/} directory. (The
13421 @file{src/} directory is below the top level of your Emacs directory.)
13424 To build this @file{TAGS} file, go to the top level of your Emacs
13425 source directory and run the compile command @code{make tags}:
13428 M-x compile RET make tags RET
13432 (The @code{make tags} command works well with the GNU Emacs sources,
13433 as well as with some other source packages.)
13435 For more information, see @ref{Tags, , Tag Tables, emacs, The GNU Emacs
13438 @node Regexp Review
13441 Here is a brief summary of some recently introduced functions.
13445 Repeatedly evaluate the body of the expression so long as the first
13446 element of the body tests true. Then return @code{nil}. (The
13447 expression is evaluated only for its side effects.)
13456 (insert (format "foo is %d.\n" foo))
13457 (setq foo (1- foo))))
13459 @result{} foo is 2.
13466 (The @code{insert} function inserts its arguments at point; the
13467 @code{format} function returns a string formatted from its arguments
13468 the way @code{message} formats its arguments; @code{\n} produces a new
13471 @item re-search-forward
13472 Search for a pattern, and if the pattern is found, move point to rest
13476 Takes four arguments, like @code{search-forward}:
13480 A regular expression that specifies the pattern to search for.
13481 (Remember to put quotation marks around this argument!)
13484 Optionally, the limit of the search.
13487 Optionally, what to do if the search fails, return @code{nil} or an
13491 Optionally, how many times to repeat the search; if negative, the
13492 search goes backwards.
13496 Bind some variables locally to particular values,
13497 and then evaluate the remaining arguments, returning the value of the
13498 last one. While binding the local variables, use the local values of
13499 variables bound earlier, if any.
13508 (message "`bar' is %d." bar))
13509 @result{} `bar' is 21.
13513 @item match-beginning
13514 Return the position of the start of the text found by the last regular
13518 Return @code{t} for true if the text after point matches the argument,
13519 which should be a regular expression.
13522 Return @code{t} for true if point is at the end of the accessible part
13523 of a buffer. The end of the accessible part is the end of the buffer
13524 if the buffer is not narrowed; it is the end of the narrowed part if
13525 the buffer is narrowed.
13529 @node re-search Exercises
13530 @section Exercises with @code{re-search-forward}
13534 Write a function to search for a regular expression that matches two
13535 or more blank lines in sequence.
13538 Write a function to search for duplicated words, such as `the the'.
13539 @xref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
13540 Manual}, for information on how to write a regexp (a regular
13541 expression) to match a string that is composed of two identical
13542 halves. You can devise several regexps; some are better than others.
13543 The function I use is described in an appendix, along with several
13544 regexps. @xref{the-the, , @code{the-the} Duplicated Words Function}.
13547 @node Counting Words
13548 @chapter Counting via Repetition and Regexps
13549 @cindex Repetition for word counting
13550 @cindex Regular expressions for word counting
13552 Repetition and regular expression searches are powerful tools that you
13553 often use when you write code in Emacs Lisp. This chapter illustrates
13554 the use of regular expression searches through the construction of
13555 word count commands using @code{while} loops and recursion.
13558 * Why Count Words::
13559 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
13560 * recursive-count-words:: Start with case of no words in region.
13561 * Counting Exercise::
13565 @node Why Count Words
13566 @unnumberedsec Counting words
13569 The standard Emacs distribution contains functions for counting the
13570 number of lines and words within a region.
13572 Certain types of writing ask you to count words. Thus, if you write
13573 an essay, you may be limited to 800 words; if you write a novel, you
13574 may discipline yourself to write 1000 words a day. It seems odd, but
13575 for a long time, Emacs lacked a word count command. Perhaps people used
13576 Emacs mostly for code or types of documentation that did not require
13577 word counts; or perhaps they restricted themselves to the operating
13578 system word count command, @code{wc}. Alternatively, people may have
13579 followed the publishers' convention and computed a word count by
13580 dividing the number of characters in a document by five.
13582 There are many ways to implement a command to count words. Here are
13583 some examples, which you may wish to compare with the standard Emacs
13584 command, @code{count-words-region}.
13586 @node @value{COUNT-WORDS}
13587 @section The @code{@value{COUNT-WORDS}} Function
13588 @findex @value{COUNT-WORDS}
13590 A word count command could count words in a line, paragraph, region,
13591 or buffer. What should the command cover? You could design the
13592 command to count the number of words in a complete buffer. However,
13593 the Emacs tradition encourages flexibility---you may want to count
13594 words in just a section, rather than all of a buffer. So it makes
13595 more sense to design the command to count the number of words in a
13596 region. Once you have a command to count words in a region, you can,
13597 if you wish, count words in a whole buffer by marking it with
13598 @w{@kbd{C-x h}} (@code{mark-whole-buffer}).
13600 Clearly, counting words is a repetitive act: starting from the
13601 beginning of the region, you count the first word, then the second
13602 word, then the third word, and so on, until you reach the end of the
13603 region. This means that word counting is ideally suited to recursion
13604 or to a @code{while} loop.
13607 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
13608 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
13612 @node Design @value{COUNT-WORDS}
13613 @unnumberedsubsec Designing @code{@value{COUNT-WORDS}}
13616 First, we will implement the word count command with a @code{while}
13617 loop, then with recursion. The command will, of course, be
13621 The template for an interactive function definition is, as always:
13625 (defun @var{name-of-function} (@var{argument-list})
13626 "@var{documentation}@dots{}"
13627 (@var{interactive-expression}@dots{})
13632 What we need to do is fill in the slots.
13634 The name of the function should be self-explanatory and similar to the
13635 existing @code{count-lines-region} name. This makes the name easier
13636 to remember. @code{count-words-region} is the obvious choice. Since
13637 that name is now used for the standard Emacs command to count words, we
13638 will name our implementation @code{@value{COUNT-WORDS}}.
13640 The function counts words within a region. This means that the
13641 argument list must contain symbols that are bound to the two
13642 positions, the beginning and end of the region. These two positions
13643 can be called @samp{beginning} and @samp{end} respectively. The first
13644 line of the documentation should be a single sentence, since that is
13645 all that is printed as documentation by a command such as
13646 @code{apropos}. The interactive expression will be of the form
13647 @samp{(interactive "r")}, since that will cause Emacs to pass the
13648 beginning and end of the region to the function's argument list. All
13651 The body of the function needs to be written to do three tasks:
13652 first, to set up conditions under which the @code{while} loop can
13653 count words, second, to run the @code{while} loop, and third, to send
13654 a message to the user.
13656 When a user calls @code{@value{COUNT-WORDS}}, point may be at the
13657 beginning or the end of the region. However, the counting process
13658 must start at the beginning of the region. This means we will want
13659 to put point there if it is not already there. Executing
13660 @code{(goto-char beginning)} ensures this. Of course, we will want to
13661 return point to its expected position when the function finishes its
13662 work. For this reason, the body must be enclosed in a
13663 @code{save-excursion} expression.
13665 The central part of the body of the function consists of a
13666 @code{while} loop in which one expression jumps point forward word by
13667 word, and another expression counts those jumps. The true-or-false-test
13668 of the @code{while} loop should test true so long as point should jump
13669 forward, and false when point is at the end of the region.
13671 We could use @code{(forward-word 1)} as the expression for moving point
13672 forward word by word, but it is easier to see what Emacs identifies as a
13673 `word' if we use a regular expression search.
13675 A regular expression search that finds the pattern for which it is
13676 searching leaves point after the last character matched. This means
13677 that a succession of successful word searches will move point forward
13680 As a practical matter, we want the regular expression search to jump
13681 over whitespace and punctuation between words as well as over the
13682 words themselves. A regexp that refuses to jump over interword
13683 whitespace would never jump more than one word! This means that
13684 the regexp should include the whitespace and punctuation that follows
13685 a word, if any, as well as the word itself. (A word may end a buffer
13686 and not have any following whitespace or punctuation, so that part of
13687 the regexp must be optional.)
13689 Thus, what we want for the regexp is a pattern defining one or more
13690 word constituent characters followed, optionally, by one or more
13691 characters that are not word constituents. The regular expression for
13699 The buffer's syntax table determines which characters are and are not
13700 word constituents. For more information about syntax,
13701 @pxref{Syntax Tables, , Syntax Tables, elisp, The GNU Emacs Lisp
13705 The search expression looks like this:
13708 (re-search-forward "\\w+\\W*")
13712 (Note that paired backslashes precede the @samp{w} and @samp{W}. A
13713 single backslash has special meaning to the Emacs Lisp interpreter.
13714 It indicates that the following character is interpreted differently
13715 than usual. For example, the two characters, @samp{\n}, stand for
13716 @samp{newline}, rather than for a backslash followed by @samp{n}. Two
13717 backslashes in a row stand for an ordinary, `unspecial' backslash, so
13718 Emacs Lisp interpreter ends of seeing a single backslash followed by a
13719 letter. So it discovers the letter is special.)
13721 We need a counter to count how many words there are; this variable
13722 must first be set to 0 and then incremented each time Emacs goes
13723 around the @code{while} loop. The incrementing expression is simply:
13726 (setq count (1+ count))
13729 Finally, we want to tell the user how many words there are in the
13730 region. The @code{message} function is intended for presenting this
13731 kind of information to the user. The message has to be phrased so
13732 that it reads properly regardless of how many words there are in the
13733 region: we don't want to say that ``there are 1 words in the region''.
13734 The conflict between singular and plural is ungrammatical. We can
13735 solve this problem by using a conditional expression that evaluates
13736 different messages depending on the number of words in the region.
13737 There are three possibilities: no words in the region, one word in the
13738 region, and more than one word. This means that the @code{cond}
13739 special form is appropriate.
13742 All this leads to the following function definition:
13746 ;;; @r{First version; has bugs!}
13747 (defun @value{COUNT-WORDS} (beginning end)
13748 "Print number of words in the region.
13749 Words are defined as at least one word-constituent
13750 character followed by at least one character that
13751 is not a word-constituent. The buffer's syntax
13752 table determines which characters these are."
13754 (message "Counting words in region ... ")
13758 ;;; @r{1. Set up appropriate conditions.}
13760 (goto-char beginning)
13765 ;;; @r{2. Run the} while @r{loop.}
13766 (while (< (point) end)
13767 (re-search-forward "\\w+\\W*")
13768 (setq count (1+ count)))
13772 ;;; @r{3. Send a message to the user.}
13773 (cond ((zerop count)
13775 "The region does NOT have any words."))
13778 "The region has 1 word."))
13781 "The region has %d words." count))))))
13786 As written, the function works, but not in all circumstances.
13788 @node Whitespace Bug
13789 @subsection The Whitespace Bug in @code{@value{COUNT-WORDS}}
13791 The @code{@value{COUNT-WORDS}} command described in the preceding
13792 section has two bugs, or rather, one bug with two manifestations.
13793 First, if you mark a region containing only whitespace in the middle
13794 of some text, the @code{@value{COUNT-WORDS}} command tells you that the
13795 region contains one word! Second, if you mark a region containing
13796 only whitespace at the end of the buffer or the accessible portion of
13797 a narrowed buffer, the command displays an error message that looks
13801 Search failed: "\\w+\\W*"
13804 If you are reading this in Info in GNU Emacs, you can test for these
13807 First, evaluate the function in the usual manner to install it.
13809 Here is a copy of the definition. Place your cursor after the closing
13810 parenthesis and type @kbd{C-x C-e} to install it.
13814 ;; @r{First version; has bugs!}
13815 (defun @value{COUNT-WORDS} (beginning end)
13816 "Print number of words in the region.
13817 Words are defined as at least one word-constituent character followed
13818 by at least one character that is not a word-constituent. The buffer's
13819 syntax table determines which characters these are."
13823 (message "Counting words in region ... ")
13827 ;;; @r{1. Set up appropriate conditions.}
13829 (goto-char beginning)
13834 ;;; @r{2. Run the} while @r{loop.}
13835 (while (< (point) end)
13836 (re-search-forward "\\w+\\W*")
13837 (setq count (1+ count)))
13841 ;;; @r{3. Send a message to the user.}
13842 (cond ((zerop count)
13843 (message "The region does NOT have any words."))
13844 ((= 1 count) (message "The region has 1 word."))
13845 (t (message "The region has %d words." count))))))
13851 If you wish, you can also install this keybinding by evaluating it:
13854 (global-set-key "\C-c=" '@value{COUNT-WORDS})
13857 To conduct the first test, set mark and point to the beginning and end
13858 of the following line and then type @kbd{C-c =} (or @kbd{M-x
13859 @value{COUNT-WORDS}} if you have not bound @kbd{C-c =}):
13866 Emacs will tell you, correctly, that the region has three words.
13868 Repeat the test, but place mark at the beginning of the line and place
13869 point just @emph{before} the word @samp{one}. Again type the command
13870 @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}). Emacs should tell you
13871 that the region has no words, since it is composed only of the
13872 whitespace at the beginning of the line. But instead Emacs tells you
13873 that the region has one word!
13875 For the third test, copy the sample line to the end of the
13876 @file{*scratch*} buffer and then type several spaces at the end of the
13877 line. Place mark right after the word @samp{three} and point at the
13878 end of line. (The end of the line will be the end of the buffer.)
13879 Type @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}) as you did before.
13880 Again, Emacs should tell you that the region has no words, since it is
13881 composed only of the whitespace at the end of the line. Instead,
13882 Emacs displays an error message saying @samp{Search failed}.
13884 The two bugs stem from the same problem.
13886 Consider the first manifestation of the bug, in which the command
13887 tells you that the whitespace at the beginning of the line contains
13888 one word. What happens is this: The @code{M-x @value{COUNT-WORDS}}
13889 command moves point to the beginning of the region. The @code{while}
13890 tests whether the value of point is smaller than the value of
13891 @code{end}, which it is. Consequently, the regular expression search
13892 looks for and finds the first word. It leaves point after the word.
13893 @code{count} is set to one. The @code{while} loop repeats; but this
13894 time the value of point is larger than the value of @code{end}, the
13895 loop is exited; and the function displays a message saying the number
13896 of words in the region is one. In brief, the regular expression
13897 search looks for and finds the word even though it is outside
13900 In the second manifestation of the bug, the region is whitespace at
13901 the end of the buffer. Emacs says @samp{Search failed}. What happens
13902 is that the true-or-false-test in the @code{while} loop tests true, so
13903 the search expression is executed. But since there are no more words
13904 in the buffer, the search fails.
13906 In both manifestations of the bug, the search extends or attempts to
13907 extend outside of the region.
13909 The solution is to limit the search to the region---this is a fairly
13910 simple action, but as you may have come to expect, it is not quite as
13911 simple as you might think.
13913 As we have seen, the @code{re-search-forward} function takes a search
13914 pattern as its first argument. But in addition to this first,
13915 mandatory argument, it accepts three optional arguments. The optional
13916 second argument bounds the search. The optional third argument, if
13917 @code{t}, causes the function to return @code{nil} rather than signal
13918 an error if the search fails. The optional fourth argument is a
13919 repeat count. (In Emacs, you can see a function's documentation by
13920 typing @kbd{C-h f}, the name of the function, and then @key{RET}.)
13922 In the @code{@value{COUNT-WORDS}} definition, the value of the end of
13923 the region is held by the variable @code{end} which is passed as an
13924 argument to the function. Thus, we can add @code{end} as an argument
13925 to the regular expression search expression:
13928 (re-search-forward "\\w+\\W*" end)
13931 However, if you make only this change to the @code{@value{COUNT-WORDS}}
13932 definition and then test the new version of the definition on a
13933 stretch of whitespace, you will receive an error message saying
13934 @samp{Search failed}.
13936 What happens is this: the search is limited to the region, and fails
13937 as you expect because there are no word-constituent characters in the
13938 region. Since it fails, we receive an error message. But we do not
13939 want to receive an error message in this case; we want to receive the
13940 message that "The region does NOT have any words."
13942 The solution to this problem is to provide @code{re-search-forward}
13943 with a third argument of @code{t}, which causes the function to return
13944 @code{nil} rather than signal an error if the search fails.
13946 However, if you make this change and try it, you will see the message
13947 ``Counting words in region ... '' and @dots{} you will keep on seeing
13948 that message @dots{}, until you type @kbd{C-g} (@code{keyboard-quit}).
13950 Here is what happens: the search is limited to the region, as before,
13951 and it fails because there are no word-constituent characters in the
13952 region, as expected. Consequently, the @code{re-search-forward}
13953 expression returns @code{nil}. It does nothing else. In particular,
13954 it does not move point, which it does as a side effect if it finds the
13955 search target. After the @code{re-search-forward} expression returns
13956 @code{nil}, the next expression in the @code{while} loop is evaluated.
13957 This expression increments the count. Then the loop repeats. The
13958 true-or-false-test tests true because the value of point is still less
13959 than the value of end, since the @code{re-search-forward} expression
13960 did not move point. @dots{} and the cycle repeats @dots{}
13962 The @code{@value{COUNT-WORDS}} definition requires yet another
13963 modification, to cause the true-or-false-test of the @code{while} loop
13964 to test false if the search fails. Put another way, there are two
13965 conditions that must be satisfied in the true-or-false-test before the
13966 word count variable is incremented: point must still be within the
13967 region and the search expression must have found a word to count.
13969 Since both the first condition and the second condition must be true
13970 together, the two expressions, the region test and the search
13971 expression, can be joined with an @code{and} special form and embedded in
13972 the @code{while} loop as the true-or-false-test, like this:
13975 (and (< (point) end) (re-search-forward "\\w+\\W*" end t))
13978 @c colon in printed section title causes problem in Info cross reference
13979 @c also trouble with an overfull hbox
13982 (For information about @code{and}, see
13983 @ref{kill-new function, , The @code{kill-new} function}.)
13987 (@xref{kill-new function, , The @code{kill-new} function}, for
13988 information about @code{and}.)
13991 The @code{re-search-forward} expression returns @code{t} if the search
13992 succeeds and as a side effect moves point. Consequently, as words are
13993 found, point is moved through the region. When the search expression
13994 fails to find another word, or when point reaches the end of the
13995 region, the true-or-false-test tests false, the @code{while} loop
13996 exits, and the @code{@value{COUNT-WORDS}} function displays one or
13997 other of its messages.
13999 After incorporating these final changes, the @code{@value{COUNT-WORDS}}
14000 works without bugs (or at least, without bugs that I have found!).
14001 Here is what it looks like:
14005 ;;; @r{Final version:} @code{while}
14006 (defun @value{COUNT-WORDS} (beginning end)
14007 "Print number of words in the region."
14009 (message "Counting words in region ... ")
14013 ;;; @r{1. Set up appropriate conditions.}
14016 (goto-char beginning)
14020 ;;; @r{2. Run the} while @r{loop.}
14021 (while (and (< (point) end)
14022 (re-search-forward "\\w+\\W*" end t))
14023 (setq count (1+ count)))
14027 ;;; @r{3. Send a message to the user.}
14028 (cond ((zerop count)
14030 "The region does NOT have any words."))
14033 "The region has 1 word."))
14036 "The region has %d words." count))))))
14040 @node recursive-count-words
14041 @section Count Words Recursively
14042 @cindex Count words recursively
14043 @cindex Recursively counting words
14044 @cindex Words, counted recursively
14046 You can write the function for counting words recursively as well as
14047 with a @code{while} loop. Let's see how this is done.
14049 First, we need to recognize that the @code{@value{COUNT-WORDS}}
14050 function has three jobs: it sets up the appropriate conditions for
14051 counting to occur; it counts the words in the region; and it sends a
14052 message to the user telling how many words there are.
14054 If we write a single recursive function to do everything, we will
14055 receive a message for every recursive call. If the region contains 13
14056 words, we will receive thirteen messages, one right after the other.
14057 We don't want this! Instead, we must write two functions to do the
14058 job, one of which (the recursive function) will be used inside of the
14059 other. One function will set up the conditions and display the
14060 message; the other will return the word count.
14062 Let us start with the function that causes the message to be displayed.
14063 We can continue to call this @code{@value{COUNT-WORDS}}.
14065 This is the function that the user will call. It will be interactive.
14066 Indeed, it will be similar to our previous versions of this
14067 function, except that it will call @code{recursive-count-words} to
14068 determine how many words are in the region.
14071 We can readily construct a template for this function, based on our
14076 ;; @r{Recursive version; uses regular expression search}
14077 (defun @value{COUNT-WORDS} (beginning end)
14078 "@var{documentation}@dots{}"
14079 (@var{interactive-expression}@dots{})
14083 ;;; @r{1. Set up appropriate conditions.}
14084 (@var{explanatory message})
14085 (@var{set-up functions}@dots{}
14089 ;;; @r{2. Count the words.}
14090 @var{recursive call}
14094 ;;; @r{3. Send a message to the user.}
14095 @var{message providing word count}))
14099 The definition looks straightforward, except that somehow the count
14100 returned by the recursive call must be passed to the message
14101 displaying the word count. A little thought suggests that this can be
14102 done by making use of a @code{let} expression: we can bind a variable
14103 in the varlist of a @code{let} expression to the number of words in
14104 the region, as returned by the recursive call; and then the
14105 @code{cond} expression, using binding, can display the value to the
14108 Often, one thinks of the binding within a @code{let} expression as
14109 somehow secondary to the `primary' work of a function. But in this
14110 case, what you might consider the `primary' job of the function,
14111 counting words, is done within the @code{let} expression.
14114 Using @code{let}, the function definition looks like this:
14118 (defun @value{COUNT-WORDS} (beginning end)
14119 "Print number of words in the region."
14124 ;;; @r{1. Set up appropriate conditions.}
14125 (message "Counting words in region ... ")
14127 (goto-char beginning)
14131 ;;; @r{2. Count the words.}
14132 (let ((count (recursive-count-words end)))
14136 ;;; @r{3. Send a message to the user.}
14137 (cond ((zerop count)
14139 "The region does NOT have any words."))
14142 "The region has 1 word."))
14145 "The region has %d words." count))))))
14149 Next, we need to write the recursive counting function.
14151 A recursive function has at least three parts: the `do-again-test', the
14152 `next-step-expression', and the recursive call.
14154 The do-again-test determines whether the function will or will not be
14155 called again. Since we are counting words in a region and can use a
14156 function that moves point forward for every word, the do-again-test
14157 can check whether point is still within the region. The do-again-test
14158 should find the value of point and determine whether point is before,
14159 at, or after the value of the end of the region. We can use the
14160 @code{point} function to locate point. Clearly, we must pass the
14161 value of the end of the region to the recursive counting function as an
14164 In addition, the do-again-test should also test whether the search finds a
14165 word. If it does not, the function should not call itself again.
14167 The next-step-expression changes a value so that when the recursive
14168 function is supposed to stop calling itself, it stops. More
14169 precisely, the next-step-expression changes a value so that at the
14170 right time, the do-again-test stops the recursive function from
14171 calling itself again. In this case, the next-step-expression can be
14172 the expression that moves point forward, word by word.
14174 The third part of a recursive function is the recursive call.
14176 Somewhere, also, we also need a part that does the `work' of the
14177 function, a part that does the counting. A vital part!
14180 But already, we have an outline of the recursive counting function:
14184 (defun recursive-count-words (region-end)
14185 "@var{documentation}@dots{}"
14186 @var{do-again-test}
14187 @var{next-step-expression}
14188 @var{recursive call})
14192 Now we need to fill in the slots. Let's start with the simplest cases
14193 first: if point is at or beyond the end of the region, there cannot
14194 be any words in the region, so the function should return zero.
14195 Likewise, if the search fails, there are no words to count, so the
14196 function should return zero.
14198 On the other hand, if point is within the region and the search
14199 succeeds, the function should call itself again.
14202 Thus, the do-again-test should look like this:
14206 (and (< (point) region-end)
14207 (re-search-forward "\\w+\\W*" region-end t))
14211 Note that the search expression is part of the do-again-test---the
14212 function returns @code{t} if its search succeeds and @code{nil} if it
14213 fails. (@xref{Whitespace Bug, , The Whitespace Bug in
14214 @code{@value{COUNT-WORDS}}}, for an explanation of how
14215 @code{re-search-forward} works.)
14217 The do-again-test is the true-or-false test of an @code{if} clause.
14218 Clearly, if the do-again-test succeeds, the then-part of the @code{if}
14219 clause should call the function again; but if it fails, the else-part
14220 should return zero since either point is outside the region or the
14221 search failed because there were no words to find.
14223 But before considering the recursive call, we need to consider the
14224 next-step-expression. What is it? Interestingly, it is the search
14225 part of the do-again-test.
14227 In addition to returning @code{t} or @code{nil} for the
14228 do-again-test, @code{re-search-forward} moves point forward as a side
14229 effect of a successful search. This is the action that changes the
14230 value of point so that the recursive function stops calling itself
14231 when point completes its movement through the region. Consequently,
14232 the @code{re-search-forward} expression is the next-step-expression.
14235 In outline, then, the body of the @code{recursive-count-words}
14236 function looks like this:
14240 (if @var{do-again-test-and-next-step-combined}
14242 @var{recursive-call-returning-count}
14248 How to incorporate the mechanism that counts?
14250 If you are not used to writing recursive functions, a question like
14251 this can be troublesome. But it can and should be approached
14254 We know that the counting mechanism should be associated in some way
14255 with the recursive call. Indeed, since the next-step-expression moves
14256 point forward by one word, and since a recursive call is made for
14257 each word, the counting mechanism must be an expression that adds one
14258 to the value returned by a call to @code{recursive-count-words}.
14261 Consider several cases:
14265 If there are two words in the region, the function should return
14266 a value resulting from adding one to the value returned when it counts
14267 the first word, plus the number returned when it counts the remaining
14268 words in the region, which in this case is one.
14271 If there is one word in the region, the function should return
14272 a value resulting from adding one to the value returned when it counts
14273 that word, plus the number returned when it counts the remaining
14274 words in the region, which in this case is zero.
14277 If there are no words in the region, the function should return zero.
14280 From the sketch we can see that the else-part of the @code{if} returns
14281 zero for the case of no words. This means that the then-part of the
14282 @code{if} must return a value resulting from adding one to the value
14283 returned from a count of the remaining words.
14286 The expression will look like this, where @code{1+} is a function that
14287 adds one to its argument.
14290 (1+ (recursive-count-words region-end))
14294 The whole @code{recursive-count-words} function will then look like
14299 (defun recursive-count-words (region-end)
14300 "@var{documentation}@dots{}"
14302 ;;; @r{1. do-again-test}
14303 (if (and (< (point) region-end)
14304 (re-search-forward "\\w+\\W*" region-end t))
14308 ;;; @r{2. then-part: the recursive call}
14309 (1+ (recursive-count-words region-end))
14311 ;;; @r{3. else-part}
14317 Let's examine how this works:
14319 If there are no words in the region, the else part of the @code{if}
14320 expression is evaluated and consequently the function returns zero.
14322 If there is one word in the region, the value of point is less than
14323 the value of @code{region-end} and the search succeeds. In this case,
14324 the true-or-false-test of the @code{if} expression tests true, and the
14325 then-part of the @code{if} expression is evaluated. The counting
14326 expression is evaluated. This expression returns a value (which will
14327 be the value returned by the whole function) that is the sum of one
14328 added to the value returned by a recursive call.
14330 Meanwhile, the next-step-expression has caused point to jump over the
14331 first (and in this case only) word in the region. This means that
14332 when @code{(recursive-count-words region-end)} is evaluated a second
14333 time, as a result of the recursive call, the value of point will be
14334 equal to or greater than the value of region end. So this time,
14335 @code{recursive-count-words} will return zero. The zero will be added
14336 to one, and the original evaluation of @code{recursive-count-words}
14337 will return one plus zero, which is one, which is the correct amount.
14339 Clearly, if there are two words in the region, the first call to
14340 @code{recursive-count-words} returns one added to the value returned
14341 by calling @code{recursive-count-words} on a region containing the
14342 remaining word---that is, it adds one to one, producing two, which is
14343 the correct amount.
14345 Similarly, if there are three words in the region, the first call to
14346 @code{recursive-count-words} returns one added to the value returned
14347 by calling @code{recursive-count-words} on a region containing the
14348 remaining two words---and so on and so on.
14352 With full documentation the two functions look like this:
14356 The recursive function:
14358 @findex recursive-count-words
14361 (defun recursive-count-words (region-end)
14362 "Number of words between point and REGION-END."
14366 ;;; @r{1. do-again-test}
14367 (if (and (< (point) region-end)
14368 (re-search-forward "\\w+\\W*" region-end t))
14372 ;;; @r{2. then-part: the recursive call}
14373 (1+ (recursive-count-words region-end))
14375 ;;; @r{3. else-part}
14386 ;;; @r{Recursive version}
14387 (defun @value{COUNT-WORDS} (beginning end)
14388 "Print number of words in the region.
14392 Words are defined as at least one word-constituent
14393 character followed by at least one character that is
14394 not a word-constituent. The buffer's syntax table
14395 determines which characters these are."
14399 (message "Counting words in region ... ")
14401 (goto-char beginning)
14402 (let ((count (recursive-count-words end)))
14405 (cond ((zerop count)
14407 "The region does NOT have any words."))
14411 (message "The region has 1 word."))
14414 "The region has %d words." count))))))
14418 @node Counting Exercise
14419 @section Exercise: Counting Punctuation
14421 Using a @code{while} loop, write a function to count the number of
14422 punctuation marks in a region---period, comma, semicolon, colon,
14423 exclamation mark, and question mark. Do the same using recursion.
14425 @node Words in a defun
14426 @chapter Counting Words in a @code{defun}
14427 @cindex Counting words in a @code{defun}
14428 @cindex Word counting in a @code{defun}
14430 Our next project is to count the number of words in a function
14431 definition. Clearly, this can be done using some variant of
14432 @code{@value{COUNT-WORDS}}. @xref{Counting Words, , Counting via
14433 Repetition and Regexps}. If we are just going to count the words in
14434 one definition, it is easy enough to mark the definition with the
14435 @kbd{C-M-h} (@code{mark-defun}) command, and then call
14436 @code{@value{COUNT-WORDS}}.
14438 However, I am more ambitious: I want to count the words and symbols in
14439 every definition in the Emacs sources and then print a graph that
14440 shows how many functions there are of each length: how many contain 40
14441 to 49 words or symbols, how many contain 50 to 59 words or symbols,
14442 and so on. I have often been curious how long a typical function is,
14443 and this will tell.
14446 * Divide and Conquer::
14447 * Words and Symbols:: What to count?
14448 * Syntax:: What constitutes a word or symbol?
14449 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
14450 * Several defuns:: Counting several defuns in a file.
14451 * Find a File:: Do you want to look at a file?
14452 * lengths-list-file:: A list of the lengths of many definitions.
14453 * Several files:: Counting in definitions in different files.
14454 * Several files recursively:: Recursively counting in different files.
14455 * Prepare the data:: Prepare the data for display in a graph.
14459 @node Divide and Conquer
14460 @unnumberedsec Divide and Conquer
14463 Described in one phrase, the histogram project is daunting; but
14464 divided into numerous small steps, each of which we can take one at a
14465 time, the project becomes less fearsome. Let us consider what the
14470 First, write a function to count the words in one definition. This
14471 includes the problem of handling symbols as well as words.
14474 Second, write a function to list the numbers of words in each function
14475 in a file. This function can use the @code{count-words-in-defun}
14479 Third, write a function to list the numbers of words in each function
14480 in each of several files. This entails automatically finding the
14481 various files, switching to them, and counting the words in the
14482 definitions within them.
14485 Fourth, write a function to convert the list of numbers that we
14486 created in step three to a form that will be suitable for printing as
14490 Fifth, write a function to print the results as a graph.
14493 This is quite a project! But if we take each step slowly, it will not
14496 @node Words and Symbols
14497 @section What to Count?
14498 @cindex Words and symbols in defun
14500 When we first start thinking about how to count the words in a
14501 function definition, the first question is (or ought to be) what are
14502 we going to count? When we speak of `words' with respect to a Lisp
14503 function definition, we are actually speaking, in large part, of
14504 `symbols'. For example, the following @code{multiply-by-seven}
14505 function contains the five symbols @code{defun},
14506 @code{multiply-by-seven}, @code{number}, @code{*}, and @code{7}. In
14507 addition, in the documentation string, it contains the four words
14508 @samp{Multiply}, @samp{NUMBER}, @samp{by}, and @samp{seven}. The
14509 symbol @samp{number} is repeated, so the definition contains a total
14510 of ten words and symbols.
14514 (defun multiply-by-seven (number)
14515 "Multiply NUMBER by seven."
14521 However, if we mark the @code{multiply-by-seven} definition with
14522 @kbd{C-M-h} (@code{mark-defun}), and then call
14523 @code{@value{COUNT-WORDS}} on it, we will find that
14524 @code{@value{COUNT-WORDS}} claims the definition has eleven words, not
14525 ten! Something is wrong!
14527 The problem is twofold: @code{@value{COUNT-WORDS}} does not count the
14528 @samp{*} as a word, and it counts the single symbol,
14529 @code{multiply-by-seven}, as containing three words. The hyphens are
14530 treated as if they were interword spaces rather than intraword
14531 connectors: @samp{multiply-by-seven} is counted as if it were written
14532 @samp{multiply by seven}.
14534 The cause of this confusion is the regular expression search within
14535 the @code{@value{COUNT-WORDS}} definition that moves point forward word
14536 by word. In the canonical version of @code{@value{COUNT-WORDS}}, the
14544 This regular expression is a pattern defining one or more word
14545 constituent characters possibly followed by one or more characters
14546 that are not word constituents. What is meant by `word constituent
14547 characters' brings us to the issue of syntax, which is worth a section
14551 @section What Constitutes a Word or Symbol?
14552 @cindex Syntax categories and tables
14554 Emacs treats different characters as belonging to different
14555 @dfn{syntax categories}. For example, the regular expression,
14556 @samp{\\w+}, is a pattern specifying one or more @emph{word
14557 constituent} characters. Word constituent characters are members of
14558 one syntax category. Other syntax categories include the class of
14559 punctuation characters, such as the period and the comma, and the
14560 class of whitespace characters, such as the blank space and the tab
14561 character. (For more information, @pxref{Syntax Tables, , Syntax
14562 Tables, elisp, The GNU Emacs Lisp Reference Manual}.)
14564 Syntax tables specify which characters belong to which categories.
14565 Usually, a hyphen is not specified as a `word constituent character'.
14566 Instead, it is specified as being in the `class of characters that are
14567 part of symbol names but not words.' This means that the
14568 @code{@value{COUNT-WORDS}} function treats it in the same way it treats
14569 an interword white space, which is why @code{@value{COUNT-WORDS}}
14570 counts @samp{multiply-by-seven} as three words.
14572 There are two ways to cause Emacs to count @samp{multiply-by-seven} as
14573 one symbol: modify the syntax table or modify the regular expression.
14575 We could redefine a hyphen as a word constituent character by
14576 modifying the syntax table that Emacs keeps for each mode. This
14577 action would serve our purpose, except that a hyphen is merely the
14578 most common character within symbols that is not typically a word
14579 constituent character; there are others, too.
14581 Alternatively, we can redefine the regexp used in the
14582 @code{@value{COUNT-WORDS}} definition so as to include symbols. This
14583 procedure has the merit of clarity, but the task is a little tricky.
14586 The first part is simple enough: the pattern must match ``at least one
14587 character that is a word or symbol constituent''. Thus:
14590 "\\(\\w\\|\\s_\\)+"
14594 The @samp{\\(} is the first part of the grouping construct that
14595 includes the @samp{\\w} and the @samp{\\s_} as alternatives, separated
14596 by the @samp{\\|}. The @samp{\\w} matches any word-constituent
14597 character and the @samp{\\s_} matches any character that is part of a
14598 symbol name but not a word-constituent character. The @samp{+}
14599 following the group indicates that the word or symbol constituent
14600 characters must be matched at least once.
14602 However, the second part of the regexp is more difficult to design.
14603 What we want is to follow the first part with ``optionally one or more
14604 characters that are not constituents of a word or symbol''. At first,
14605 I thought I could define this with the following:
14608 "\\(\\W\\|\\S_\\)*"
14612 The upper case @samp{W} and @samp{S} match characters that are
14613 @emph{not} word or symbol constituents. Unfortunately, this
14614 expression matches any character that is either not a word constituent
14615 or not a symbol constituent. This matches any character!
14617 I then noticed that every word or symbol in my test region was
14618 followed by white space (blank space, tab, or newline). So I tried
14619 placing a pattern to match one or more blank spaces after the pattern
14620 for one or more word or symbol constituents. This failed, too. Words
14621 and symbols are often separated by whitespace, but in actual code
14622 parentheses may follow symbols and punctuation may follow words. So
14623 finally, I designed a pattern in which the word or symbol constituents
14624 are followed optionally by characters that are not white space and
14625 then followed optionally by white space.
14628 Here is the full regular expression:
14631 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14634 @node count-words-in-defun
14635 @section The @code{count-words-in-defun} Function
14636 @cindex Counting words in a @code{defun}
14638 We have seen that there are several ways to write a
14639 @code{count-words-region} function. To write a
14640 @code{count-words-in-defun}, we need merely adapt one of these
14643 The version that uses a @code{while} loop is easy to understand, so I
14644 am going to adapt that. Because @code{count-words-in-defun} will be
14645 part of a more complex program, it need not be interactive and it need
14646 not display a message but just return the count. These considerations
14647 simplify the definition a little.
14649 On the other hand, @code{count-words-in-defun} will be used within a
14650 buffer that contains function definitions. Consequently, it is
14651 reasonable to ask that the function determine whether it is called
14652 when point is within a function definition, and if it is, to return
14653 the count for that definition. This adds complexity to the
14654 definition, but saves us from needing to pass arguments to the
14658 These considerations lead us to prepare the following template:
14662 (defun count-words-in-defun ()
14663 "@var{documentation}@dots{}"
14664 (@var{set up}@dots{}
14665 (@var{while loop}@dots{})
14666 @var{return count})
14671 As usual, our job is to fill in the slots.
14675 We are presuming that this function will be called within a buffer
14676 containing function definitions. Point will either be within a
14677 function definition or not. For @code{count-words-in-defun} to work,
14678 point must move to the beginning of the definition, a counter must
14679 start at zero, and the counting loop must stop when point reaches the
14680 end of the definition.
14682 The @code{beginning-of-defun} function searches backwards for an
14683 opening delimiter such as a @samp{(} at the beginning of a line, and
14684 moves point to that position, or else to the limit of the search. In
14685 practice, this means that @code{beginning-of-defun} moves point to the
14686 beginning of an enclosing or preceding function definition, or else to
14687 the beginning of the buffer. We can use @code{beginning-of-defun} to
14688 place point where we wish to start.
14690 The @code{while} loop requires a counter to keep track of the words or
14691 symbols being counted. A @code{let} expression can be used to create
14692 a local variable for this purpose, and bind it to an initial value of zero.
14694 The @code{end-of-defun} function works like @code{beginning-of-defun}
14695 except that it moves point to the end of the definition.
14696 @code{end-of-defun} can be used as part of an expression that
14697 determines the position of the end of the definition.
14699 The set up for @code{count-words-in-defun} takes shape rapidly: first
14700 we move point to the beginning of the definition, then we create a
14701 local variable to hold the count, and finally, we record the position
14702 of the end of the definition so the @code{while} loop will know when to stop
14706 The code looks like this:
14710 (beginning-of-defun)
14712 (end (save-excursion (end-of-defun) (point))))
14717 The code is simple. The only slight complication is likely to concern
14718 @code{end}: it is bound to the position of the end of the definition
14719 by a @code{save-excursion} expression that returns the value of point
14720 after @code{end-of-defun} temporarily moves it to the end of the
14723 The second part of the @code{count-words-in-defun}, after the set up,
14724 is the @code{while} loop.
14726 The loop must contain an expression that jumps point forward word by
14727 word and symbol by symbol, and another expression that counts the
14728 jumps. The true-or-false-test for the @code{while} loop should test
14729 true so long as point should jump forward, and false when point is at
14730 the end of the definition. We have already redefined the regular
14731 expression for this, so the loop is straightforward:
14735 (while (and (< (point) end)
14737 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t))
14738 (setq count (1+ count)))
14742 The third part of the function definition returns the count of words
14743 and symbols. This part is the last expression within the body of the
14744 @code{let} expression, and can be, very simply, the local variable
14745 @code{count}, which when evaluated returns the count.
14748 Put together, the @code{count-words-in-defun} definition looks like this:
14750 @findex count-words-in-defun
14753 (defun count-words-in-defun ()
14754 "Return the number of words and symbols in a defun."
14755 (beginning-of-defun)
14757 (end (save-excursion (end-of-defun) (point))))
14761 (and (< (point) end)
14763 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14765 (setq count (1+ count)))
14770 How to test this? The function is not interactive, but it is easy to
14771 put a wrapper around the function to make it interactive; we can use
14772 almost the same code as for the recursive version of
14773 @code{@value{COUNT-WORDS}}:
14777 ;;; @r{Interactive version.}
14778 (defun count-words-defun ()
14779 "Number of words and symbols in a function definition."
14782 "Counting words and symbols in function definition ... ")
14785 (let ((count (count-words-in-defun)))
14789 "The definition does NOT have any words or symbols."))
14794 "The definition has 1 word or symbol."))
14797 "The definition has %d words or symbols." count)))))
14803 Let's re-use @kbd{C-c =} as a convenient keybinding:
14806 (global-set-key "\C-c=" 'count-words-defun)
14809 Now we can try out @code{count-words-defun}: install both
14810 @code{count-words-in-defun} and @code{count-words-defun}, and set the
14811 keybinding, and then place the cursor within the following definition:
14815 (defun multiply-by-seven (number)
14816 "Multiply NUMBER by seven."
14823 Success! The definition has 10 words and symbols.
14825 The next problem is to count the numbers of words and symbols in
14826 several definitions within a single file.
14828 @node Several defuns
14829 @section Count Several @code{defuns} Within a File
14831 A file such as @file{simple.el} may have a hundred or more function
14832 definitions within it. Our long term goal is to collect statistics on
14833 many files, but as a first step, our immediate goal is to collect
14834 statistics on one file.
14836 The information will be a series of numbers, each number being the
14837 length of a function definition. We can store the numbers in a list.
14839 We know that we will want to incorporate the information regarding one
14840 file with information about many other files; this means that the
14841 function for counting definition lengths within one file need only
14842 return the list of lengths. It need not and should not display any
14845 The word count commands contain one expression to jump point forward
14846 word by word and another expression to count the jumps. The function
14847 to return the lengths of definitions can be designed to work the same
14848 way, with one expression to jump point forward definition by
14849 definition and another expression to construct the lengths' list.
14851 This statement of the problem makes it elementary to write the
14852 function definition. Clearly, we will start the count at the
14853 beginning of the file, so the first command will be @code{(goto-char
14854 (point-min))}. Next, we start the @code{while} loop; and the
14855 true-or-false test of the loop can be a regular expression search for
14856 the next function definition---so long as the search succeeds, point
14857 is moved forward and then the body of the loop is evaluated. The body
14858 needs an expression that constructs the lengths' list. @code{cons},
14859 the list construction command, can be used to create the list. That
14860 is almost all there is to it.
14863 Here is what this fragment of code looks like:
14867 (goto-char (point-min))
14868 (while (re-search-forward "^(defun" nil t)
14870 (cons (count-words-in-defun) lengths-list)))
14874 What we have left out is the mechanism for finding the file that
14875 contains the function definitions.
14877 In previous examples, we either used this, the Info file, or we
14878 switched back and forth to some other buffer, such as the
14879 @file{*scratch*} buffer.
14881 Finding a file is a new process that we have not yet discussed.
14884 @section Find a File
14885 @cindex Find a File
14887 To find a file in Emacs, you use the @kbd{C-x C-f} (@code{find-file})
14888 command. This command is almost, but not quite right for the lengths
14892 Let's look at the source for @code{find-file}:
14896 (defun find-file (filename)
14897 "Edit file FILENAME.
14898 Switch to a buffer visiting file FILENAME,
14899 creating one if none already exists."
14900 (interactive "FFind file: ")
14901 (switch-to-buffer (find-file-noselect filename)))
14906 (The most recent version of the @code{find-file} function definition
14907 permits you to specify optional wildcards to visit multiple files; that
14908 makes the definition more complex and we will not discuss it here,
14909 since it is not relevant. You can see its source using either
14910 @kbd{M-.} (@code{find-tag}) or @kbd{C-h f} (@code{describe-function}).)
14914 (defun find-file (filename &optional wildcards)
14915 "Edit file FILENAME.
14916 Switch to a buffer visiting file FILENAME,
14917 creating one if none already exists.
14918 Interactively, the default if you just type RET is the current directory,
14919 but the visited file name is available through the minibuffer history:
14920 type M-n to pull it into the minibuffer.
14922 Interactively, or if WILDCARDS is non-nil in a call from Lisp,
14923 expand wildcards (if any) and visit multiple files. You can
14924 suppress wildcard expansion by setting `find-file-wildcards' to nil.
14926 To visit a file without any kind of conversion and without
14927 automatically choosing a major mode, use \\[find-file-literally]."
14928 (interactive (find-file-read-args "Find file: " nil))
14929 (let ((value (find-file-noselect filename nil nil wildcards)))
14931 (mapcar 'switch-to-buffer (nreverse value))
14932 (switch-to-buffer value))))
14935 The definition I am showing possesses short but complete documentation
14936 and an interactive specification that prompts you for a file name when
14937 you use the command interactively. The body of the definition
14938 contains two functions, @code{find-file-noselect} and
14939 @code{switch-to-buffer}.
14941 According to its documentation as shown by @kbd{C-h f} (the
14942 @code{describe-function} command), the @code{find-file-noselect}
14943 function reads the named file into a buffer and returns the buffer.
14944 (Its most recent version includes an optional wildcards argument,
14945 too, as well as another to read a file literally and an other you
14946 suppress warning messages. These optional arguments are irrelevant.)
14948 However, the @code{find-file-noselect} function does not select the
14949 buffer in which it puts the file. Emacs does not switch its attention
14950 (or yours if you are using @code{find-file-noselect}) to the selected
14951 buffer. That is what @code{switch-to-buffer} does: it switches the
14952 buffer to which Emacs attention is directed; and it switches the
14953 buffer displayed in the window to the new buffer. We have discussed
14954 buffer switching elsewhere. (@xref{Switching Buffers}.)
14956 In this histogram project, we do not need to display each file on the
14957 screen as the program determines the length of each definition within
14958 it. Instead of employing @code{switch-to-buffer}, we can work with
14959 @code{set-buffer}, which redirects the attention of the computer
14960 program to a different buffer but does not redisplay it on the screen.
14961 So instead of calling on @code{find-file} to do the job, we must write
14962 our own expression.
14964 The task is easy: use @code{find-file-noselect} and @code{set-buffer}.
14966 @node lengths-list-file
14967 @section @code{lengths-list-file} in Detail
14969 The core of the @code{lengths-list-file} function is a @code{while}
14970 loop containing a function to move point forward `defun by defun' and
14971 a function to count the number of words and symbols in each defun.
14972 This core must be surrounded by functions that do various other tasks,
14973 including finding the file, and ensuring that point starts out at the
14974 beginning of the file. The function definition looks like this:
14975 @findex lengths-list-file
14979 (defun lengths-list-file (filename)
14980 "Return list of definitions' lengths within FILE.
14981 The returned list is a list of numbers.
14982 Each number is the number of words or
14983 symbols in one function definition."
14986 (message "Working on `%s' ... " filename)
14988 (let ((buffer (find-file-noselect filename))
14990 (set-buffer buffer)
14991 (setq buffer-read-only t)
14993 (goto-char (point-min))
14994 (while (re-search-forward "^(defun" nil t)
14996 (cons (count-words-in-defun) lengths-list)))
14997 (kill-buffer buffer)
15003 The function is passed one argument, the name of the file on which it
15004 will work. It has four lines of documentation, but no interactive
15005 specification. Since people worry that a computer is broken if they
15006 don't see anything going on, the first line of the body is a
15009 The next line contains a @code{save-excursion} that returns Emacs's
15010 attention to the current buffer when the function completes. This is
15011 useful in case you embed this function in another function that
15012 presumes point is restored to the original buffer.
15014 In the varlist of the @code{let} expression, Emacs finds the file and
15015 binds the local variable @code{buffer} to the buffer containing the
15016 file. At the same time, Emacs creates @code{lengths-list} as a local
15019 Next, Emacs switches its attention to the buffer.
15021 In the following line, Emacs makes the buffer read-only. Ideally,
15022 this line is not necessary. None of the functions for counting words
15023 and symbols in a function definition should change the buffer.
15024 Besides, the buffer is not going to be saved, even if it were changed.
15025 This line is entirely the consequence of great, perhaps excessive,
15026 caution. The reason for the caution is that this function and those
15027 it calls work on the sources for Emacs and it is inconvenient if they
15028 are inadvertently modified. It goes without saying that I did not
15029 realize a need for this line until an experiment went awry and started
15030 to modify my Emacs source files @dots{}
15032 Next comes a call to widen the buffer if it is narrowed. This
15033 function is usually not needed---Emacs creates a fresh buffer if none
15034 already exists; but if a buffer visiting the file already exists Emacs
15035 returns that one. In this case, the buffer may be narrowed and must
15036 be widened. If we wanted to be fully `user-friendly', we would
15037 arrange to save the restriction and the location of point, but we
15040 The @code{(goto-char (point-min))} expression moves point to the
15041 beginning of the buffer.
15043 Then comes a @code{while} loop in which the `work' of the function is
15044 carried out. In the loop, Emacs determines the length of each
15045 definition and constructs a lengths' list containing the information.
15047 Emacs kills the buffer after working through it. This is to save
15048 space inside of Emacs. My version of GNU Emacs 19 contained over 300
15049 source files of interest; GNU Emacs 22 contains over a thousand source
15050 files. Another function will apply @code{lengths-list-file} to each
15053 Finally, the last expression within the @code{let} expression is the
15054 @code{lengths-list} variable; its value is returned as the value of
15055 the whole function.
15057 You can try this function by installing it in the usual fashion. Then
15058 place your cursor after the following expression and type @kbd{C-x
15059 C-e} (@code{eval-last-sexp}).
15061 @c !!! 22.1.1 lisp sources location here
15064 "/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el")
15068 (You may need to change the pathname of the file; the one here is for
15069 GNU Emacs version 22.1.1. To change the expression, copy it to
15070 the @file{*scratch*} buffer and edit it.
15074 (Also, to see the full length of the list, rather than a truncated
15075 version, you may have to evaluate the following:
15078 (custom-set-variables '(eval-expression-print-length nil))
15082 (@xref{defcustom, , Specifying Variables using @code{defcustom}}.
15083 Then evaluate the @code{lengths-list-file} expression.)
15086 The lengths' list for @file{debug.el} takes less than a second to
15087 produce and looks like this in GNU Emacs 22:
15090 (83 113 105 144 289 22 30 97 48 89 25 52 52 88 28 29 77 49 43 290 232 587)
15094 (Using my old machine, the version 19 lengths' list for @file{debug.el}
15095 took seven seconds to produce and looked like this:
15098 (75 41 80 62 20 45 44 68 45 12 34 235)
15101 (The newer version of @file{debug.el} contains more defuns than the
15102 earlier one; and my new machine is much faster than the old one.)
15104 Note that the length of the last definition in the file is first in
15107 @node Several files
15108 @section Count Words in @code{defuns} in Different Files
15110 In the previous section, we created a function that returns a list of
15111 the lengths of each definition in a file. Now, we want to define a
15112 function to return a master list of the lengths of the definitions in
15115 Working on each of a list of files is a repetitious act, so we can use
15116 either a @code{while} loop or recursion.
15119 * lengths-list-many-files:: Return a list of the lengths of defuns.
15120 * append:: Attach one list to another.
15124 @node lengths-list-many-files
15125 @unnumberedsubsec Determine the lengths of @code{defuns}
15128 The design using a @code{while} loop is routine. The argument passed
15129 the function is a list of files. As we saw earlier (@pxref{Loop
15130 Example}), you can write a @code{while} loop so that the body of the
15131 loop is evaluated if such a list contains elements, but to exit the
15132 loop if the list is empty. For this design to work, the body of the
15133 loop must contain an expression that shortens the list each time the
15134 body is evaluated, so that eventually the list is empty. The usual
15135 technique is to set the value of the list to the value of the @sc{cdr}
15136 of the list each time the body is evaluated.
15139 The template looks like this:
15143 (while @var{test-whether-list-is-empty}
15145 @var{set-list-to-cdr-of-list})
15149 Also, we remember that a @code{while} loop returns @code{nil} (the
15150 result of evaluating the true-or-false-test), not the result of any
15151 evaluation within its body. (The evaluations within the body of the
15152 loop are done for their side effects.) However, the expression that
15153 sets the lengths' list is part of the body---and that is the value
15154 that we want returned by the function as a whole. To do this, we
15155 enclose the @code{while} loop within a @code{let} expression, and
15156 arrange that the last element of the @code{let} expression contains
15157 the value of the lengths' list. (@xref{Incrementing Example, , Loop
15158 Example with an Incrementing Counter}.)
15160 @findex lengths-list-many-files
15162 These considerations lead us directly to the function itself:
15166 ;;; @r{Use @code{while} loop.}
15167 (defun lengths-list-many-files (list-of-files)
15168 "Return list of lengths of defuns in LIST-OF-FILES."
15171 (let (lengths-list)
15173 ;;; @r{true-or-false-test}
15174 (while list-of-files
15179 ;;; @r{Generate a lengths' list.}
15181 (expand-file-name (car list-of-files)))))
15185 ;;; @r{Make files' list shorter.}
15186 (setq list-of-files (cdr list-of-files)))
15188 ;;; @r{Return final value of lengths' list.}
15193 @code{expand-file-name} is a built-in function that converts a file
15194 name to the absolute, long, path name form. The function employs the
15195 name of the directory in which the function is called.
15197 @c !!! 22.1.1 lisp sources location here
15199 Thus, if @code{expand-file-name} is called on @code{debug.el} when
15200 Emacs is visiting the
15201 @file{/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/} directory,
15211 @c !!! 22.1.1 lisp sources location here
15213 /usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el
15216 The only other new element of this function definition is the as yet
15217 unstudied function @code{append}, which merits a short section for
15221 @subsection The @code{append} Function
15224 The @code{append} function attaches one list to another. Thus,
15227 (append '(1 2 3 4) '(5 6 7 8))
15238 This is exactly how we want to attach two lengths' lists produced by
15239 @code{lengths-list-file} to each other. The results contrast with
15243 (cons '(1 2 3 4) '(5 6 7 8))
15248 which constructs a new list in which the first argument to @code{cons}
15249 becomes the first element of the new list:
15252 ((1 2 3 4) 5 6 7 8)
15255 @node Several files recursively
15256 @section Recursively Count Words in Different Files
15258 Besides a @code{while} loop, you can work on each of a list of files
15259 with recursion. A recursive version of @code{lengths-list-many-files}
15260 is short and simple.
15262 The recursive function has the usual parts: the `do-again-test', the
15263 `next-step-expression', and the recursive call. The `do-again-test'
15264 determines whether the function should call itself again, which it
15265 will do if the @code{list-of-files} contains any remaining elements;
15266 the `next-step-expression' resets the @code{list-of-files} to the
15267 @sc{cdr} of itself, so eventually the list will be empty; and the
15268 recursive call calls itself on the shorter list. The complete
15269 function is shorter than this description!
15270 @findex recursive-lengths-list-many-files
15274 (defun recursive-lengths-list-many-files (list-of-files)
15275 "Return list of lengths of each defun in LIST-OF-FILES."
15276 (if list-of-files ; @r{do-again-test}
15279 (expand-file-name (car list-of-files)))
15280 (recursive-lengths-list-many-files
15281 (cdr list-of-files)))))
15286 In a sentence, the function returns the lengths' list for the first of
15287 the @code{list-of-files} appended to the result of calling itself on
15288 the rest of the @code{list-of-files}.
15290 Here is a test of @code{recursive-lengths-list-many-files}, along with
15291 the results of running @code{lengths-list-file} on each of the files
15294 Install @code{recursive-lengths-list-many-files} and
15295 @code{lengths-list-file}, if necessary, and then evaluate the
15296 following expressions. You may need to change the files' pathnames;
15297 those here work when this Info file and the Emacs sources are located
15298 in their customary places. To change the expressions, copy them to
15299 the @file{*scratch*} buffer, edit them, and then evaluate them.
15301 The results are shown after the @samp{@result{}}. (These results are
15302 for files from Emacs version 22.1.1; files from other versions of
15303 Emacs may produce different results.)
15305 @c !!! 22.1.1 lisp sources location here
15308 (cd "/usr/local/share/emacs/22.1.1/")
15310 (lengths-list-file "./lisp/macros.el")
15311 @result{} (283 263 480 90)
15315 (lengths-list-file "./lisp/mail/mailalias.el")
15316 @result{} (38 32 29 95 178 180 321 218 324)
15320 (lengths-list-file "./lisp/makesum.el")
15325 (recursive-lengths-list-many-files
15326 '("./lisp/macros.el"
15327 "./lisp/mail/mailalias.el"
15328 "./lisp/makesum.el"))
15329 @result{} (283 263 480 90 38 32 29 95 178 180 321 218 324 85 181)
15333 The @code{recursive-lengths-list-many-files} function produces the
15336 The next step is to prepare the data in the list for display in a graph.
15338 @node Prepare the data
15339 @section Prepare the Data for Display in a Graph
15341 The @code{recursive-lengths-list-many-files} function returns a list
15342 of numbers. Each number records the length of a function definition.
15343 What we need to do now is transform this data into a list of numbers
15344 suitable for generating a graph. The new list will tell how many
15345 functions definitions contain less than 10 words and
15346 symbols, how many contain between 10 and 19 words and symbols, how
15347 many contain between 20 and 29 words and symbols, and so on.
15349 In brief, we need to go through the lengths' list produced by the
15350 @code{recursive-lengths-list-many-files} function and count the number
15351 of defuns within each range of lengths, and produce a list of those
15355 * Data for Display in Detail::
15356 * Sorting:: Sorting lists.
15357 * Files List:: Making a list of files.
15358 * Counting function definitions::
15362 @node Data for Display in Detail
15363 @unnumberedsubsec The Data for Display in Detail
15366 Based on what we have done before, we can readily foresee that it
15367 should not be too hard to write a function that `@sc{cdr}s' down the
15368 lengths' list, looks at each element, determines which length range it
15369 is in, and increments a counter for that range.
15371 However, before beginning to write such a function, we should consider
15372 the advantages of sorting the lengths' list first, so the numbers are
15373 ordered from smallest to largest. First, sorting will make it easier
15374 to count the numbers in each range, since two adjacent numbers will
15375 either be in the same length range or in adjacent ranges. Second, by
15376 inspecting a sorted list, we can discover the highest and lowest
15377 number, and thereby determine the largest and smallest length range
15381 @subsection Sorting Lists
15384 Emacs contains a function to sort lists, called (as you might guess)
15385 @code{sort}. The @code{sort} function takes two arguments, the list
15386 to be sorted, and a predicate that determines whether the first of
15387 two list elements is ``less'' than the second.
15389 As we saw earlier (@pxref{Wrong Type of Argument, , Using the Wrong
15390 Type Object as an Argument}), a predicate is a function that
15391 determines whether some property is true or false. The @code{sort}
15392 function will reorder a list according to whatever property the
15393 predicate uses; this means that @code{sort} can be used to sort
15394 non-numeric lists by non-numeric criteria---it can, for example,
15395 alphabetize a list.
15398 The @code{<} function is used when sorting a numeric list. For example,
15401 (sort '(4 8 21 17 33 7 21 7) '<)
15409 (4 7 7 8 17 21 21 33)
15413 (Note that in this example, both the arguments are quoted so that the
15414 symbols are not evaluated before being passed to @code{sort} as
15417 Sorting the list returned by the
15418 @code{recursive-lengths-list-many-files} function is straightforward;
15419 it uses the @code{<} function:
15423 In GNU Emacs 22, eval
15425 (cd "/usr/local/share/emacs/22.0.50/")
15427 (recursive-lengths-list-many-files
15428 '("./lisp/macros.el"
15429 "./lisp/mail/mailalias.el"
15430 "./lisp/makesum.el"))
15438 (recursive-lengths-list-many-files
15439 '("./lisp/macros.el"
15440 "./lisp/mailalias.el"
15441 "./lisp/makesum.el"))
15451 (29 32 38 85 90 95 178 180 181 218 263 283 321 324 480)
15455 (Note that in this example, the first argument to @code{sort} is not
15456 quoted, since the expression must be evaluated so as to produce the
15457 list that is passed to @code{sort}.)
15460 @subsection Making a List of Files
15462 The @code{recursive-lengths-list-many-files} function requires a list
15463 of files as its argument. For our test examples, we constructed such
15464 a list by hand; but the Emacs Lisp source directory is too large for
15465 us to do for that. Instead, we will write a function to do the job
15466 for us. In this function, we will use both a @code{while} loop and a
15469 @findex directory-files
15470 We did not have to write a function like this for older versions of
15471 GNU Emacs, since they placed all the @samp{.el} files in one
15472 directory. Instead, we were able to use the @code{directory-files}
15473 function, which lists the names of files that match a specified
15474 pattern within a single directory.
15476 However, recent versions of Emacs place Emacs Lisp files in
15477 sub-directories of the top level @file{lisp} directory. This
15478 re-arrangement eases navigation. For example, all the mail related
15479 files are in a @file{lisp} sub-directory called @file{mail}. But at
15480 the same time, this arrangement forces us to create a file listing
15481 function that descends into the sub-directories.
15483 @findex files-in-below-directory
15484 We can create this function, called @code{files-in-below-directory},
15485 using familiar functions such as @code{car}, @code{nthcdr}, and
15486 @code{substring} in conjunction with an existing function called
15487 @code{directory-files-and-attributes}. This latter function not only
15488 lists all the filenames in a directory, including the names
15489 of sub-directories, but also their attributes.
15491 To restate our goal: to create a function that will enable us
15492 to feed filenames to @code{recursive-lengths-list-many-files}
15493 as a list that looks like this (but with more elements):
15497 ("./lisp/macros.el"
15498 "./lisp/mail/rmail.el"
15499 "./lisp/makesum.el")
15503 The @code{directory-files-and-attributes} function returns a list of
15504 lists. Each of the lists within the main list consists of 13
15505 elements. The first element is a string that contains the name of the
15506 file---which, in GNU/Linux, may be a `directory file', that is to
15507 say, a file with the special attributes of a directory. The second
15508 element of the list is @code{t} for a directory, a string
15509 for symbolic link (the string is the name linked to), or @code{nil}.
15511 For example, the first @samp{.el} file in the @file{lisp/} directory
15512 is @file{abbrev.el}. Its name is
15513 @file{/usr/local/share/emacs/22.1.1/lisp/abbrev.el} and it is not a
15514 directory or a symbolic link.
15517 This is how @code{directory-files-and-attributes} lists that file and
15529 (20615 27034 579989 697000)
15531 (20615 26327 734791 805000)
15543 On the other hand, @file{mail/} is a directory within the @file{lisp/}
15544 directory. The beginning of its listing looks like this:
15555 (To learn about the different attributes, look at the documentation of
15556 @code{file-attributes}. Bear in mind that the @code{file-attributes}
15557 function does not list the filename, so its first element is
15558 @code{directory-files-and-attributes}'s second element.)
15560 We will want our new function, @code{files-in-below-directory}, to
15561 list the @samp{.el} files in the directory it is told to check, and in
15562 any directories below that directory.
15564 This gives us a hint on how to construct
15565 @code{files-in-below-directory}: within a directory, the function
15566 should add @samp{.el} filenames to a list; and if, within a directory,
15567 the function comes upon a sub-directory, it should go into that
15568 sub-directory and repeat its actions.
15570 However, we should note that every directory contains a name that
15571 refers to itself, called @file{.}, (``dot'') and a name that refers to
15572 its parent directory, called @file{..} (``double dot''). (In
15573 @file{/}, the root directory, @file{..} refers to itself, since
15574 @file{/} has no parent.) Clearly, we do not want our
15575 @code{files-in-below-directory} function to enter those directories,
15576 since they always lead us, directly or indirectly, to the current
15579 Consequently, our @code{files-in-below-directory} function must do
15584 Check to see whether it is looking at a filename that ends in
15585 @samp{.el}; and if so, add its name to a list.
15588 Check to see whether it is looking at a filename that is the name of a
15589 directory; and if so,
15593 Check to see whether it is looking at @file{.} or @file{..}; and if
15597 Or else, go into that directory and repeat the process.
15601 Let's write a function definition to do these tasks. We will use a
15602 @code{while} loop to move from one filename to another within a
15603 directory, checking what needs to be done; and we will use a recursive
15604 call to repeat the actions on each sub-directory. The recursive
15605 pattern is `accumulate'
15606 (@pxref{Accumulate}),
15607 using @code{append} as the combiner.
15610 (directory-files "/usr/local/src/emacs/lisp/" t "\\.el$")
15611 (shell-command "find /usr/local/src/emacs/lisp/ -name '*.el'")
15613 (directory-files "/usr/local/share/emacs/22.1.1/lisp/" t "\\.el$")
15614 (shell-command "find /usr/local/share/emacs/22.1.1/lisp/ -name '*.el'")
15617 @c /usr/local/share/emacs/22.1.1/lisp/
15620 Here is the function:
15624 (defun files-in-below-directory (directory)
15625 "List the .el files in DIRECTORY and in its sub-directories."
15626 ;; Although the function will be used non-interactively,
15627 ;; it will be easier to test if we make it interactive.
15628 ;; The directory will have a name such as
15629 ;; "/usr/local/share/emacs/22.1.1/lisp/"
15630 (interactive "DDirectory name: ")
15633 (let (el-files-list
15634 (current-directory-list
15635 (directory-files-and-attributes directory t)))
15636 ;; while we are in the current directory
15637 (while current-directory-list
15641 ;; check to see whether filename ends in `.el'
15642 ;; and if so, append its name to a list.
15643 ((equal ".el" (substring (car (car current-directory-list)) -3))
15644 (setq el-files-list
15645 (cons (car (car current-directory-list)) el-files-list)))
15648 ;; check whether filename is that of a directory
15649 ((eq t (car (cdr (car current-directory-list))))
15650 ;; decide whether to skip or recurse
15653 (substring (car (car current-directory-list)) -1))
15654 ;; then do nothing since filename is that of
15655 ;; current directory or parent, "." or ".."
15659 ;; else descend into the directory and repeat the process
15660 (setq el-files-list
15662 (files-in-below-directory
15663 (car (car current-directory-list)))
15665 ;; move to the next filename in the list; this also
15666 ;; shortens the list so the while loop eventually comes to an end
15667 (setq current-directory-list (cdr current-directory-list)))
15668 ;; return the filenames
15673 @c (files-in-below-directory "/usr/local/src/emacs/lisp/")
15674 @c (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15676 The @code{files-in-below-directory} @code{directory-files} function
15677 takes one argument, the name of a directory.
15680 Thus, on my system,
15682 @c (length (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15684 @c !!! 22.1.1 lisp sources location here
15688 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/"))
15693 tells me that in and below my Lisp sources directory are 1031
15696 @code{files-in-below-directory} returns a list in reverse alphabetical
15697 order. An expression to sort the list in alphabetical order looks
15703 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15710 "Test how long it takes to find lengths of all sorted elisp defuns."
15711 (insert "\n" (current-time-string) "\n")
15714 (recursive-lengths-list-many-files
15715 (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15717 (insert (format "%s" (current-time-string))))
15720 @node Counting function definitions
15721 @subsection Counting function definitions
15723 Our immediate goal is to generate a list that tells us how many
15724 function definitions contain fewer than 10 words and symbols, how many
15725 contain between 10 and 19 words and symbols, how many contain between
15726 20 and 29 words and symbols, and so on.
15728 With a sorted list of numbers, this is easy: count how many elements
15729 of the list are smaller than 10, then, after moving past the numbers
15730 just counted, count how many are smaller than 20, then, after moving
15731 past the numbers just counted, count how many are smaller than 30, and
15732 so on. Each of the numbers, 10, 20, 30, 40, and the like, is one
15733 larger than the top of that range. We can call the list of such
15734 numbers the @code{top-of-ranges} list.
15737 If we wished, we could generate this list automatically, but it is
15738 simpler to write a list manually. Here it is:
15739 @vindex top-of-ranges
15743 (defvar top-of-ranges
15746 110 120 130 140 150
15747 160 170 180 190 200
15748 210 220 230 240 250
15749 260 270 280 290 300)
15750 "List specifying ranges for `defuns-per-range'.")
15754 To change the ranges, we edit this list.
15756 Next, we need to write the function that creates the list of the
15757 number of definitions within each range. Clearly, this function must
15758 take the @code{sorted-lengths} and the @code{top-of-ranges} lists
15761 The @code{defuns-per-range} function must do two things again and
15762 again: it must count the number of definitions within a range
15763 specified by the current top-of-range value; and it must shift to the
15764 next higher value in the @code{top-of-ranges} list after counting the
15765 number of definitions in the current range. Since each of these
15766 actions is repetitive, we can use @code{while} loops for the job.
15767 One loop counts the number of definitions in the range defined by the
15768 current top-of-range value, and the other loop selects each of the
15769 top-of-range values in turn.
15771 Several entries of the @code{sorted-lengths} list are counted for each
15772 range; this means that the loop for the @code{sorted-lengths} list
15773 will be inside the loop for the @code{top-of-ranges} list, like a
15774 small gear inside a big gear.
15776 The inner loop counts the number of definitions within the range. It
15777 is a simple counting loop of the type we have seen before.
15778 (@xref{Incrementing Loop, , A loop with an incrementing counter}.)
15779 The true-or-false test of the loop tests whether the value from the
15780 @code{sorted-lengths} list is smaller than the current value of the
15781 top of the range. If it is, the function increments the counter and
15782 tests the next value from the @code{sorted-lengths} list.
15785 The inner loop looks like this:
15789 (while @var{length-element-smaller-than-top-of-range}
15790 (setq number-within-range (1+ number-within-range))
15791 (setq sorted-lengths (cdr sorted-lengths)))
15795 The outer loop must start with the lowest value of the
15796 @code{top-of-ranges} list, and then be set to each of the succeeding
15797 higher values in turn. This can be done with a loop like this:
15801 (while top-of-ranges
15802 @var{body-of-loop}@dots{}
15803 (setq top-of-ranges (cdr top-of-ranges)))
15808 Put together, the two loops look like this:
15812 (while top-of-ranges
15814 ;; @r{Count the number of elements within the current range.}
15815 (while @var{length-element-smaller-than-top-of-range}
15816 (setq number-within-range (1+ number-within-range))
15817 (setq sorted-lengths (cdr sorted-lengths)))
15819 ;; @r{Move to next range.}
15820 (setq top-of-ranges (cdr top-of-ranges)))
15824 In addition, in each circuit of the outer loop, Emacs should record
15825 the number of definitions within that range (the value of
15826 @code{number-within-range}) in a list. We can use @code{cons} for
15827 this purpose. (@xref{cons, , @code{cons}}.)
15829 The @code{cons} function works fine, except that the list it
15830 constructs will contain the number of definitions for the highest
15831 range at its beginning and the number of definitions for the lowest
15832 range at its end. This is because @code{cons} attaches new elements
15833 of the list to the beginning of the list, and since the two loops are
15834 working their way through the lengths' list from the lower end first,
15835 the @code{defuns-per-range-list} will end up largest number first.
15836 But we will want to print our graph with smallest values first and the
15837 larger later. The solution is to reverse the order of the
15838 @code{defuns-per-range-list}. We can do this using the
15839 @code{nreverse} function, which reverses the order of a list.
15846 (nreverse '(1 2 3 4))
15857 Note that the @code{nreverse} function is ``destructive''---that is,
15858 it changes the list to which it is applied; this contrasts with the
15859 @code{car} and @code{cdr} functions, which are non-destructive. In
15860 this case, we do not want the original @code{defuns-per-range-list},
15861 so it does not matter that it is destroyed. (The @code{reverse}
15862 function provides a reversed copy of a list, leaving the original list
15867 Put all together, the @code{defuns-per-range} looks like this:
15871 (defun defuns-per-range (sorted-lengths top-of-ranges)
15872 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
15873 (let ((top-of-range (car top-of-ranges))
15874 (number-within-range 0)
15875 defuns-per-range-list)
15880 (while top-of-ranges
15886 ;; @r{Need number for numeric test.}
15887 (car sorted-lengths)
15888 (< (car sorted-lengths) top-of-range))
15892 ;; @r{Count number of definitions within current range.}
15893 (setq number-within-range (1+ number-within-range))
15894 (setq sorted-lengths (cdr sorted-lengths)))
15896 ;; @r{Exit inner loop but remain within outer loop.}
15900 (setq defuns-per-range-list
15901 (cons number-within-range defuns-per-range-list))
15902 (setq number-within-range 0) ; @r{Reset count to zero.}
15906 ;; @r{Move to next range.}
15907 (setq top-of-ranges (cdr top-of-ranges))
15908 ;; @r{Specify next top of range value.}
15909 (setq top-of-range (car top-of-ranges)))
15913 ;; @r{Exit outer loop and count the number of defuns larger than}
15914 ;; @r{ the largest top-of-range value.}
15915 (setq defuns-per-range-list
15917 (length sorted-lengths)
15918 defuns-per-range-list))
15922 ;; @r{Return a list of the number of definitions within each range,}
15923 ;; @r{ smallest to largest.}
15924 (nreverse defuns-per-range-list)))
15930 The function is straightforward except for one subtle feature. The
15931 true-or-false test of the inner loop looks like this:
15935 (and (car sorted-lengths)
15936 (< (car sorted-lengths) top-of-range))
15942 instead of like this:
15945 (< (car sorted-lengths) top-of-range)
15948 The purpose of the test is to determine whether the first item in the
15949 @code{sorted-lengths} list is less than the value of the top of the
15952 The simple version of the test works fine unless the
15953 @code{sorted-lengths} list has a @code{nil} value. In that case, the
15954 @code{(car sorted-lengths)} expression function returns
15955 @code{nil}. The @code{<} function cannot compare a number to
15956 @code{nil}, which is an empty list, so Emacs signals an error and
15957 stops the function from attempting to continue to execute.
15959 The @code{sorted-lengths} list always becomes @code{nil} when the
15960 counter reaches the end of the list. This means that any attempt to
15961 use the @code{defuns-per-range} function with the simple version of
15962 the test will fail.
15964 We solve the problem by using the @code{(car sorted-lengths)}
15965 expression in conjunction with the @code{and} expression. The
15966 @code{(car sorted-lengths)} expression returns a non-@code{nil}
15967 value so long as the list has at least one number within it, but
15968 returns @code{nil} if the list is empty. The @code{and} expression
15969 first evaluates the @code{(car sorted-lengths)} expression, and
15970 if it is @code{nil}, returns false @emph{without} evaluating the
15971 @code{<} expression. But if the @code{(car sorted-lengths)}
15972 expression returns a non-@code{nil} value, the @code{and} expression
15973 evaluates the @code{<} expression, and returns that value as the value
15974 of the @code{and} expression.
15976 @c colon in printed section title causes problem in Info cross reference
15977 This way, we avoid an error.
15980 (For information about @code{and}, see
15981 @ref{kill-new function, , The @code{kill-new} function}.)
15985 (@xref{kill-new function, , The @code{kill-new} function}, for
15986 information about @code{and}.)
15989 Here is a short test of the @code{defuns-per-range} function. First,
15990 evaluate the expression that binds (a shortened)
15991 @code{top-of-ranges} list to the list of values, then evaluate the
15992 expression for binding the @code{sorted-lengths} list, and then
15993 evaluate the @code{defuns-per-range} function.
15997 ;; @r{(Shorter list than we will use later.)}
15998 (setq top-of-ranges
15999 '(110 120 130 140 150
16000 160 170 180 190 200))
16002 (setq sorted-lengths
16003 '(85 86 110 116 122 129 154 176 179 200 265 300 300))
16005 (defuns-per-range sorted-lengths top-of-ranges)
16011 The list returned looks like this:
16014 (2 2 2 0 0 1 0 2 0 0 4)
16018 Indeed, there are two elements of the @code{sorted-lengths} list
16019 smaller than 110, two elements between 110 and 119, two elements
16020 between 120 and 129, and so on. There are four elements with a value
16023 @c The next step is to turn this numbers' list into a graph.
16024 @node Readying a Graph
16025 @chapter Readying a Graph
16026 @cindex Readying a graph
16027 @cindex Graph prototype
16028 @cindex Prototype graph
16029 @cindex Body of graph
16031 Our goal is to construct a graph showing the numbers of function
16032 definitions of various lengths in the Emacs lisp sources.
16034 As a practical matter, if you were creating a graph, you would
16035 probably use a program such as @code{gnuplot} to do the job.
16036 (@code{gnuplot} is nicely integrated into GNU Emacs.) In this case,
16037 however, we create one from scratch, and in the process we will
16038 re-acquaint ourselves with some of what we learned before and learn
16041 In this chapter, we will first write a simple graph printing function.
16042 This first definition will be a @dfn{prototype}, a rapidly written
16043 function that enables us to reconnoiter this unknown graph-making
16044 territory. We will discover dragons, or find that they are myth.
16045 After scouting the terrain, we will feel more confident and enhance
16046 the function to label the axes automatically.
16049 * Columns of a graph::
16050 * graph-body-print:: How to print the body of a graph.
16051 * recursive-graph-body-print::
16053 * Line Graph Exercise::
16057 @node Columns of a graph
16058 @unnumberedsec Printing the Columns of a Graph
16061 Since Emacs is designed to be flexible and work with all kinds of
16062 terminals, including character-only terminals, the graph will need to
16063 be made from one of the `typewriter' symbols. An asterisk will do; as
16064 we enhance the graph-printing function, we can make the choice of
16065 symbol a user option.
16067 We can call this function @code{graph-body-print}; it will take a
16068 @code{numbers-list} as its only argument. At this stage, we will not
16069 label the graph, but only print its body.
16071 The @code{graph-body-print} function inserts a vertical column of
16072 asterisks for each element in the @code{numbers-list}. The height of
16073 each line is determined by the value of that element of the
16074 @code{numbers-list}.
16076 Inserting columns is a repetitive act; that means that this function can
16077 be written either with a @code{while} loop or recursively.
16079 Our first challenge is to discover how to print a column of asterisks.
16080 Usually, in Emacs, we print characters onto a screen horizontally,
16081 line by line, by typing. We have two routes we can follow: write our
16082 own column-insertion function or discover whether one exists in Emacs.
16084 To see whether there is one in Emacs, we can use the @kbd{M-x apropos}
16085 command. This command is like the @kbd{C-h a} (@code{command-apropos})
16086 command, except that the latter finds only those functions that are
16087 commands. The @kbd{M-x apropos} command lists all symbols that match
16088 a regular expression, including functions that are not interactive.
16091 What we want to look for is some command that prints or inserts
16092 columns. Very likely, the name of the function will contain either
16093 the word `print' or the word `insert' or the word `column'.
16094 Therefore, we can simply type @kbd{M-x apropos RET
16095 print\|insert\|column RET} and look at the result. On my system, this
16096 command once too takes quite some time, and then produced a list of 79
16097 functions and variables. Now it does not take much time at all and
16098 produces a list of 211 functions and variables. Scanning down the
16099 list, the only function that looks as if it might do the job is
16100 @code{insert-rectangle}.
16103 Indeed, this is the function we want; its documentation says:
16108 Insert text of RECTANGLE with upper left corner at point.
16109 RECTANGLE's first line is inserted at point,
16110 its second line is inserted at a point vertically under point, etc.
16111 RECTANGLE should be a list of strings.
16112 After this command, the mark is at the upper left corner
16113 and point is at the lower right corner.
16117 We can run a quick test, to make sure it does what we expect of it.
16119 Here is the result of placing the cursor after the
16120 @code{insert-rectangle} expression and typing @kbd{C-u C-x C-e}
16121 (@code{eval-last-sexp}). The function inserts the strings
16122 @samp{"first"}, @samp{"second"}, and @samp{"third"} at and below
16123 point. Also the function returns @code{nil}.
16127 (insert-rectangle '("first" "second" "third"))first
16134 Of course, we won't be inserting the text of the
16135 @code{insert-rectangle} expression itself into the buffer in which we
16136 are making the graph, but will call the function from our program. We
16137 shall, however, have to make sure that point is in the buffer at the
16138 place where the @code{insert-rectangle} function will insert its
16141 If you are reading this in Info, you can see how this works by
16142 switching to another buffer, such as the @file{*scratch*} buffer,
16143 placing point somewhere in the buffer, typing @kbd{M-:}, typing the
16144 @code{insert-rectangle} expression into the minibuffer at the prompt,
16145 and then typing @key{RET}. This causes Emacs to evaluate the
16146 expression in the minibuffer, but to use as the value of point the
16147 position of point in the @file{*scratch*} buffer. (@kbd{M-:} is the
16148 keybinding for @code{eval-expression}. Also, @code{nil} does not
16149 appear in the @file{*scratch*} buffer since the expression is
16150 evaluated in the minibuffer.)
16152 We find when we do this that point ends up at the end of the last
16153 inserted line---that is to say, this function moves point as a
16154 side-effect. If we were to repeat the command, with point at this
16155 position, the next insertion would be below and to the right of the
16156 previous insertion. We don't want this! If we are going to make a
16157 bar graph, the columns need to be beside each other.
16159 So we discover that each cycle of the column-inserting @code{while}
16160 loop must reposition point to the place we want it, and that place
16161 will be at the top, not the bottom, of the column. Moreover, we
16162 remember that when we print a graph, we do not expect all the columns
16163 to be the same height. This means that the top of each column may be
16164 at a different height from the previous one. We cannot simply
16165 reposition point to the same line each time, but moved over to the
16166 right---or perhaps we can@dots{}
16168 We are planning to make the columns of the bar graph out of asterisks.
16169 The number of asterisks in the column is the number specified by the
16170 current element of the @code{numbers-list}. We need to construct a
16171 list of asterisks of the right length for each call to
16172 @code{insert-rectangle}. If this list consists solely of the requisite
16173 number of asterisks, then we will have position point the right number
16174 of lines above the base for the graph to print correctly. This could
16177 Alternatively, if we can figure out some way to pass
16178 @code{insert-rectangle} a list of the same length each time, then we
16179 can place point on the same line each time, but move it over one
16180 column to the right for each new column. If we do this, however, some
16181 of the entries in the list passed to @code{insert-rectangle} must be
16182 blanks rather than asterisks. For example, if the maximum height of
16183 the graph is 5, but the height of the column is 3, then
16184 @code{insert-rectangle} requires an argument that looks like this:
16187 (" " " " "*" "*" "*")
16190 This last proposal is not so difficult, so long as we can determine
16191 the column height. There are two ways for us to specify the column
16192 height: we can arbitrarily state what it will be, which would work
16193 fine for graphs of that height; or we can search through the list of
16194 numbers and use the maximum height of the list as the maximum height
16195 of the graph. If the latter operation were difficult, then the former
16196 procedure would be easiest, but there is a function built into Emacs
16197 that determines the maximum of its arguments. We can use that
16198 function. The function is called @code{max} and it returns the
16199 largest of all its arguments, which must be numbers. Thus, for
16207 returns 7. (A corresponding function called @code{min} returns the
16208 smallest of all its arguments.)
16212 However, we cannot simply call @code{max} on the @code{numbers-list};
16213 the @code{max} function expects numbers as its argument, not a list of
16214 numbers. Thus, the following expression,
16217 (max '(3 4 6 5 7 3))
16222 produces the following error message;
16225 Wrong type of argument: number-or-marker-p, (3 4 6 5 7 3)
16229 We need a function that passes a list of arguments to a function.
16230 This function is @code{apply}. This function `applies' its first
16231 argument (a function) to its remaining arguments, the last of which
16238 (apply 'max 3 4 7 3 '(4 8 5))
16244 (Incidentally, I don't know how you would learn of this function
16245 without a book such as this. It is possible to discover other
16246 functions, like @code{search-forward} or @code{insert-rectangle}, by
16247 guessing at a part of their names and then using @code{apropos}. Even
16248 though its base in metaphor is clear---`apply' its first argument to
16249 the rest---I doubt a novice would come up with that particular word
16250 when using @code{apropos} or other aid. Of course, I could be wrong;
16251 after all, the function was first named by someone who had to invent
16254 The second and subsequent arguments to @code{apply} are optional, so
16255 we can use @code{apply} to call a function and pass the elements of a
16256 list to it, like this, which also returns 8:
16259 (apply 'max '(4 8 5))
16262 This latter way is how we will use @code{apply}. The
16263 @code{recursive-lengths-list-many-files} function returns a numbers'
16264 list to which we can apply @code{max} (we could also apply @code{max} to
16265 the sorted numbers' list; it does not matter whether the list is
16269 Hence, the operation for finding the maximum height of the graph is this:
16272 (setq max-graph-height (apply 'max numbers-list))
16275 Now we can return to the question of how to create a list of strings
16276 for a column of the graph. Told the maximum height of the graph
16277 and the number of asterisks that should appear in the column, the
16278 function should return a list of strings for the
16279 @code{insert-rectangle} command to insert.
16281 Each column is made up of asterisks or blanks. Since the function is
16282 passed the value of the height of the column and the number of
16283 asterisks in the column, the number of blanks can be found by
16284 subtracting the number of asterisks from the height of the column.
16285 Given the number of blanks and the number of asterisks, two
16286 @code{while} loops can be used to construct the list:
16290 ;;; @r{First version.}
16291 (defun column-of-graph (max-graph-height actual-height)
16292 "Return list of strings that is one column of a graph."
16293 (let ((insert-list nil)
16294 (number-of-top-blanks
16295 (- max-graph-height actual-height)))
16299 ;; @r{Fill in asterisks.}
16300 (while (> actual-height 0)
16301 (setq insert-list (cons "*" insert-list))
16302 (setq actual-height (1- actual-height)))
16306 ;; @r{Fill in blanks.}
16307 (while (> number-of-top-blanks 0)
16308 (setq insert-list (cons " " insert-list))
16309 (setq number-of-top-blanks
16310 (1- number-of-top-blanks)))
16314 ;; @r{Return whole list.}
16319 If you install this function and then evaluate the following
16320 expression you will see that it returns the list as desired:
16323 (column-of-graph 5 3)
16331 (" " " " "*" "*" "*")
16334 As written, @code{column-of-graph} contains a major flaw: the symbols
16335 used for the blank and for the marked entries in the column are
16336 `hard-coded' as a space and asterisk. This is fine for a prototype,
16337 but you, or another user, may wish to use other symbols. For example,
16338 in testing the graph function, you many want to use a period in place
16339 of the space, to make sure the point is being repositioned properly
16340 each time the @code{insert-rectangle} function is called; or you might
16341 want to substitute a @samp{+} sign or other symbol for the asterisk.
16342 You might even want to make a graph-column that is more than one
16343 display column wide. The program should be more flexible. The way to
16344 do that is to replace the blank and the asterisk with two variables
16345 that we can call @code{graph-blank} and @code{graph-symbol} and define
16346 those variables separately.
16348 Also, the documentation is not well written. These considerations
16349 lead us to the second version of the function:
16353 (defvar graph-symbol "*"
16354 "String used as symbol in graph, usually an asterisk.")
16358 (defvar graph-blank " "
16359 "String used as blank in graph, usually a blank space.
16360 graph-blank must be the same number of columns wide
16366 (For an explanation of @code{defvar}, see
16367 @ref{defvar, , Initializing a Variable with @code{defvar}}.)
16371 ;;; @r{Second version.}
16372 (defun column-of-graph (max-graph-height actual-height)
16373 "Return MAX-GRAPH-HEIGHT strings; ACTUAL-HEIGHT are graph-symbols.
16377 The graph-symbols are contiguous entries at the end
16379 The list will be inserted as one column of a graph.
16380 The strings are either graph-blank or graph-symbol."
16384 (let ((insert-list nil)
16385 (number-of-top-blanks
16386 (- max-graph-height actual-height)))
16390 ;; @r{Fill in @code{graph-symbols}.}
16391 (while (> actual-height 0)
16392 (setq insert-list (cons graph-symbol insert-list))
16393 (setq actual-height (1- actual-height)))
16397 ;; @r{Fill in @code{graph-blanks}.}
16398 (while (> number-of-top-blanks 0)
16399 (setq insert-list (cons graph-blank insert-list))
16400 (setq number-of-top-blanks
16401 (1- number-of-top-blanks)))
16403 ;; @r{Return whole list.}
16408 If we wished, we could rewrite @code{column-of-graph} a third time to
16409 provide optionally for a line graph as well as for a bar graph. This
16410 would not be hard to do. One way to think of a line graph is that it
16411 is no more than a bar graph in which the part of each bar that is
16412 below the top is blank. To construct a column for a line graph, the
16413 function first constructs a list of blanks that is one shorter than
16414 the value, then it uses @code{cons} to attach a graph symbol to the
16415 list; then it uses @code{cons} again to attach the `top blanks' to
16418 It is easy to see how to write such a function, but since we don't
16419 need it, we will not do it. But the job could be done, and if it were
16420 done, it would be done with @code{column-of-graph}. Even more
16421 important, it is worth noting that few changes would have to be made
16422 anywhere else. The enhancement, if we ever wish to make it, is
16425 Now, finally, we come to our first actual graph printing function.
16426 This prints the body of a graph, not the labels for the vertical and
16427 horizontal axes, so we can call this @code{graph-body-print}.
16429 @node graph-body-print
16430 @section The @code{graph-body-print} Function
16431 @findex graph-body-print
16433 After our preparation in the preceding section, the
16434 @code{graph-body-print} function is straightforward. The function
16435 will print column after column of asterisks and blanks, using the
16436 elements of a numbers' list to specify the number of asterisks in each
16437 column. This is a repetitive act, which means we can use a
16438 decrementing @code{while} loop or recursive function for the job. In
16439 this section, we will write the definition using a @code{while} loop.
16441 The @code{column-of-graph} function requires the height of the graph
16442 as an argument, so we should determine and record that as a local variable.
16444 This leads us to the following template for the @code{while} loop
16445 version of this function:
16449 (defun graph-body-print (numbers-list)
16450 "@var{documentation}@dots{}"
16451 (let ((height @dots{}
16456 (while numbers-list
16457 @var{insert-columns-and-reposition-point}
16458 (setq numbers-list (cdr numbers-list)))))
16463 We need to fill in the slots of the template.
16465 Clearly, we can use the @code{(apply 'max numbers-list)} expression to
16466 determine the height of the graph.
16468 The @code{while} loop will cycle through the @code{numbers-list} one
16469 element at a time. As it is shortened by the @code{(setq numbers-list
16470 (cdr numbers-list))} expression, the @sc{car} of each instance of the
16471 list is the value of the argument for @code{column-of-graph}.
16473 At each cycle of the @code{while} loop, the @code{insert-rectangle}
16474 function inserts the list returned by @code{column-of-graph}. Since
16475 the @code{insert-rectangle} function moves point to the lower right of
16476 the inserted rectangle, we need to save the location of point at the
16477 time the rectangle is inserted, move back to that position after the
16478 rectangle is inserted, and then move horizontally to the next place
16479 from which @code{insert-rectangle} is called.
16481 If the inserted columns are one character wide, as they will be if
16482 single blanks and asterisks are used, the repositioning command is
16483 simply @code{(forward-char 1)}; however, the width of a column may be
16484 greater than one. This means that the repositioning command should be
16485 written @code{(forward-char symbol-width)}. The @code{symbol-width}
16486 itself is the length of a @code{graph-blank} and can be found using
16487 the expression @code{(length graph-blank)}. The best place to bind
16488 the @code{symbol-width} variable to the value of the width of graph
16489 column is in the varlist of the @code{let} expression.
16492 These considerations lead to the following function definition:
16496 (defun graph-body-print (numbers-list)
16497 "Print a bar graph of the NUMBERS-LIST.
16498 The numbers-list consists of the Y-axis values."
16500 (let ((height (apply 'max numbers-list))
16501 (symbol-width (length graph-blank))
16506 (while numbers-list
16507 (setq from-position (point))
16509 (column-of-graph height (car numbers-list)))
16510 (goto-char from-position)
16511 (forward-char symbol-width)
16514 ;; @r{Draw graph column by column.}
16516 (setq numbers-list (cdr numbers-list)))
16519 ;; @r{Place point for X axis labels.}
16520 (forward-line height)
16527 The one unexpected expression in this function is the
16528 @w{@code{(sit-for 0)}} expression in the @code{while} loop. This
16529 expression makes the graph printing operation more interesting to
16530 watch than it would be otherwise. The expression causes Emacs to
16531 `sit' or do nothing for a zero length of time and then redraw the
16532 screen. Placed here, it causes Emacs to redraw the screen column by
16533 column. Without it, Emacs would not redraw the screen until the
16536 We can test @code{graph-body-print} with a short list of numbers.
16540 Install @code{graph-symbol}, @code{graph-blank},
16541 @code{column-of-graph}, which are in
16543 @ref{Readying a Graph, , Readying a Graph},
16546 @ref{Columns of a graph},
16548 and @code{graph-body-print}.
16552 Copy the following expression:
16555 (graph-body-print '(1 2 3 4 6 4 3 5 7 6 5 2 3))
16559 Switch to the @file{*scratch*} buffer and place the cursor where you
16560 want the graph to start.
16563 Type @kbd{M-:} (@code{eval-expression}).
16566 Yank the @code{graph-body-print} expression into the minibuffer
16567 with @kbd{C-y} (@code{yank)}.
16570 Press @key{RET} to evaluate the @code{graph-body-print} expression.
16574 Emacs will print a graph like this:
16588 @node recursive-graph-body-print
16589 @section The @code{recursive-graph-body-print} Function
16590 @findex recursive-graph-body-print
16592 The @code{graph-body-print} function may also be written recursively.
16593 The recursive solution is divided into two parts: an outside `wrapper'
16594 that uses a @code{let} expression to determine the values of several
16595 variables that need only be found once, such as the maximum height of
16596 the graph, and an inside function that is called recursively to print
16600 The `wrapper' is uncomplicated:
16604 (defun recursive-graph-body-print (numbers-list)
16605 "Print a bar graph of the NUMBERS-LIST.
16606 The numbers-list consists of the Y-axis values."
16607 (let ((height (apply 'max numbers-list))
16608 (symbol-width (length graph-blank))
16610 (recursive-graph-body-print-internal
16617 The recursive function is a little more difficult. It has four parts:
16618 the `do-again-test', the printing code, the recursive call, and the
16619 `next-step-expression'. The `do-again-test' is a @code{when}
16620 expression that determines whether the @code{numbers-list} contains
16621 any remaining elements; if it does, the function prints one column of
16622 the graph using the printing code and calls itself again. The
16623 function calls itself again according to the value produced by the
16624 `next-step-expression' which causes the call to act on a shorter
16625 version of the @code{numbers-list}.
16629 (defun recursive-graph-body-print-internal
16630 (numbers-list height symbol-width)
16631 "Print a bar graph.
16632 Used within recursive-graph-body-print function."
16637 (setq from-position (point))
16639 (column-of-graph height (car numbers-list)))
16642 (goto-char from-position)
16643 (forward-char symbol-width)
16644 (sit-for 0) ; @r{Draw graph column by column.}
16645 (recursive-graph-body-print-internal
16646 (cdr numbers-list) height symbol-width)))
16651 After installation, this expression can be tested; here is a sample:
16654 (recursive-graph-body-print '(3 2 5 6 7 5 3 4 6 4 3 2 1))
16658 Here is what @code{recursive-graph-body-print} produces:
16672 Either of these two functions, @code{graph-body-print} or
16673 @code{recursive-graph-body-print}, create the body of a graph.
16676 @section Need for Printed Axes
16678 A graph needs printed axes, so you can orient yourself. For a do-once
16679 project, it may be reasonable to draw the axes by hand using Emacs's
16680 Picture mode; but a graph drawing function may be used more than once.
16682 For this reason, I have written enhancements to the basic
16683 @code{print-graph-body} function that automatically print labels for
16684 the horizontal and vertical axes. Since the label printing functions
16685 do not contain much new material, I have placed their description in
16686 an appendix. @xref{Full Graph, , A Graph with Labeled Axes}.
16688 @node Line Graph Exercise
16691 Write a line graph version of the graph printing functions.
16693 @node Emacs Initialization
16694 @chapter Your @file{.emacs} File
16695 @cindex @file{.emacs} file
16696 @cindex Customizing your @file{.emacs} file
16697 @cindex Initialization file
16699 ``You don't have to like Emacs to like it''---this seemingly
16700 paradoxical statement is the secret of GNU Emacs. The plain, `out of
16701 the box' Emacs is a generic tool. Most people who use it, customize
16702 it to suit themselves.
16704 GNU Emacs is mostly written in Emacs Lisp; this means that by writing
16705 expressions in Emacs Lisp you can change or extend Emacs.
16708 * Default Configuration::
16709 * Site-wide Init:: You can write site-wide init files.
16710 * defcustom:: Emacs will write code for you.
16711 * Beginning init File:: How to write a @file{.emacs} init file.
16712 * Text and Auto-fill:: Automatically wrap lines.
16713 * Mail Aliases:: Use abbreviations for email addresses.
16714 * Indent Tabs Mode:: Don't use tabs with @TeX{}
16715 * Keybindings:: Create some personal keybindings.
16716 * Keymaps:: More about key binding.
16717 * Loading Files:: Load (i.e., evaluate) files automatically.
16718 * Autoload:: Make functions available.
16719 * Simple Extension:: Define a function; bind it to a key.
16720 * X11 Colors:: Colors in X.
16722 * Mode Line:: How to customize your mode line.
16726 @node Default Configuration
16727 @unnumberedsec Emacs's Default Configuration
16730 There are those who appreciate Emacs's default configuration. After
16731 all, Emacs starts you in C mode when you edit a C file, starts you in
16732 Fortran mode when you edit a Fortran file, and starts you in
16733 Fundamental mode when you edit an unadorned file. This all makes
16734 sense, if you do not know who is going to use Emacs. Who knows what a
16735 person hopes to do with an unadorned file? Fundamental mode is the
16736 right default for such a file, just as C mode is the right default for
16737 editing C code. (Enough programming languages have syntaxes
16738 that enable them to share or nearly share features, so C mode is
16739 now provided by CC mode, the `C Collection'.)
16741 But when you do know who is going to use Emacs---you,
16742 yourself---then it makes sense to customize Emacs.
16744 For example, I seldom want Fundamental mode when I edit an
16745 otherwise undistinguished file; I want Text mode. This is why I
16746 customize Emacs: so it suits me.
16748 You can customize and extend Emacs by writing or adapting a
16749 @file{~/.emacs} file. This is your personal initialization file; its
16750 contents, written in Emacs Lisp, tell Emacs what to do.@footnote{You
16751 may also add @file{.el} to @file{~/.emacs} and call it a
16752 @file{~/.emacs.el} file. In the past, you were forbidden to type the
16753 extra keystrokes that the name @file{~/.emacs.el} requires, but now
16754 you may. The new format is consistent with the Emacs Lisp file
16755 naming conventions; the old format saves typing.}
16757 A @file{~/.emacs} file contains Emacs Lisp code. You can write this
16758 code yourself; or you can use Emacs's @code{customize} feature to write
16759 the code for you. You can combine your own expressions and
16760 auto-written Customize expressions in your @file{.emacs} file.
16762 (I myself prefer to write my own expressions, except for those,
16763 particularly fonts, that I find easier to manipulate using the
16764 @code{customize} command. I combine the two methods.)
16766 Most of this chapter is about writing expressions yourself. It
16767 describes a simple @file{.emacs} file; for more information, see
16768 @ref{Init File, , The Init File, emacs, The GNU Emacs Manual}, and
16769 @ref{Init File, , The Init File, elisp, The GNU Emacs Lisp Reference
16772 @node Site-wide Init
16773 @section Site-wide Initialization Files
16775 @cindex @file{default.el} init file
16776 @cindex @file{site-init.el} init file
16777 @cindex @file{site-load.el} init file
16778 In addition to your personal initialization file, Emacs automatically
16779 loads various site-wide initialization files, if they exist. These
16780 have the same form as your @file{.emacs} file, but are loaded by
16783 Two site-wide initialization files, @file{site-load.el} and
16784 @file{site-init.el}, are loaded into Emacs and then `dumped' if a
16785 `dumped' version of Emacs is created, as is most common. (Dumped
16786 copies of Emacs load more quickly. However, once a file is loaded and
16787 dumped, a change to it does not lead to a change in Emacs unless you
16788 load it yourself or re-dump Emacs. @xref{Building Emacs, , Building
16789 Emacs, elisp, The GNU Emacs Lisp Reference Manual}, and the
16790 @file{INSTALL} file.)
16792 Three other site-wide initialization files are loaded automatically
16793 each time you start Emacs, if they exist. These are
16794 @file{site-start.el}, which is loaded @emph{before} your @file{.emacs}
16795 file, and @file{default.el}, and the terminal type file, which are both
16796 loaded @emph{after} your @file{.emacs} file.
16798 Settings and definitions in your @file{.emacs} file will overwrite
16799 conflicting settings and definitions in a @file{site-start.el} file,
16800 if it exists; but the settings and definitions in a @file{default.el}
16801 or terminal type file will overwrite those in your @file{.emacs} file.
16802 (You can prevent interference from a terminal type file by setting
16803 @code{term-file-prefix} to @code{nil}. @xref{Simple Extension, , A
16804 Simple Extension}.)
16806 @c Rewritten to avoid overfull hbox.
16807 The @file{INSTALL} file that comes in the distribution contains
16808 descriptions of the @file{site-init.el} and @file{site-load.el} files.
16810 The @file{loadup.el}, @file{startup.el}, and @file{loaddefs.el} files
16811 control loading. These files are in the @file{lisp} directory of the
16812 Emacs distribution and are worth perusing.
16814 The @file{loaddefs.el} file contains a good many suggestions as to
16815 what to put into your own @file{.emacs} file, or into a site-wide
16816 initialization file.
16819 @section Specifying Variables using @code{defcustom}
16822 You can specify variables using @code{defcustom} so that you and
16823 others can then use Emacs's @code{customize} feature to set their
16824 values. (You cannot use @code{customize} to write function
16825 definitions; but you can write @code{defuns} in your @file{.emacs}
16826 file. Indeed, you can write any Lisp expression in your @file{.emacs}
16829 The @code{customize} feature depends on the @code{defcustom} macro.
16830 Although you can use @code{defvar} or @code{setq} for variables that
16831 users set, the @code{defcustom} macro is designed for the job.
16833 You can use your knowledge of @code{defvar} for writing the
16834 first three arguments for @code{defcustom}. The first argument to
16835 @code{defcustom} is the name of the variable. The second argument is
16836 the variable's initial value, if any; and this value is set only if
16837 the value has not already been set. The third argument is the
16840 The fourth and subsequent arguments to @code{defcustom} specify types
16841 and options; these are not featured in @code{defvar}. (These
16842 arguments are optional.)
16844 Each of these arguments consists of a keyword followed by a value.
16845 Each keyword starts with the colon character @samp{:}.
16848 For example, the customizable user option variable
16849 @code{text-mode-hook} looks like this:
16853 (defcustom text-mode-hook nil
16854 "Normal hook run when entering Text mode and many related modes."
16856 :options '(turn-on-auto-fill flyspell-mode)
16862 The name of the variable is @code{text-mode-hook}; it has no default
16863 value; and its documentation string tells you what it does.
16865 The @code{:type} keyword tells Emacs the kind of data to which
16866 @code{text-mode-hook} should be set and how to display the value in a
16867 Customization buffer.
16869 The @code{:options} keyword specifies a suggested list of values for
16870 the variable. Usually, @code{:options} applies to a hook.
16871 The list is only a suggestion; it is not exclusive; a person who sets
16872 the variable may set it to other values; the list shown following the
16873 @code{:options} keyword is intended to offer convenient choices to a
16876 Finally, the @code{:group} keyword tells the Emacs Customization
16877 command in which group the variable is located. This tells where to
16880 The @code{defcustom} macro recognizes more than a dozen keywords.
16881 For more information, see @ref{Customization, , Writing Customization
16882 Definitions, elisp, The GNU Emacs Lisp Reference Manual}.
16884 Consider @code{text-mode-hook} as an example.
16886 There are two ways to customize this variable. You can use the
16887 customization command or write the appropriate expressions yourself.
16890 Using the customization command, you can type:
16897 and find that the group for editing files of data is called `data'.
16898 Enter that group. Text Mode Hook is the first member. You can click
16899 on its various options, such as @code{turn-on-auto-fill}, to set the
16900 values. After you click on the button to
16903 Save for Future Sessions
16907 Emacs will write an expression into your @file{.emacs} file.
16908 It will look like this:
16912 (custom-set-variables
16913 ;; custom-set-variables was added by Custom.
16914 ;; If you edit it by hand, you could mess it up, so be careful.
16915 ;; Your init file should contain only one such instance.
16916 ;; If there is more than one, they won't work right.
16917 '(text-mode-hook (quote (turn-on-auto-fill text-mode-hook-identify))))
16922 (The @code{text-mode-hook-identify} function tells
16923 @code{toggle-text-mode-auto-fill} which buffers are in Text mode.
16924 It comes on automatically.)
16926 The @code{custom-set-variables} function works somewhat differently
16927 than a @code{setq}. While I have never learned the differences, I
16928 modify the @code{custom-set-variables} expressions in my @file{.emacs}
16929 file by hand: I make the changes in what appears to me to be a
16930 reasonable manner and have not had any problems. Others prefer to use
16931 the Customization command and let Emacs do the work for them.
16933 Another @code{custom-set-@dots{}} function is @code{custom-set-faces}.
16934 This function sets the various font faces. Over time, I have set a
16935 considerable number of faces. Some of the time, I re-set them using
16936 @code{customize}; other times, I simply edit the
16937 @code{custom-set-faces} expression in my @file{.emacs} file itself.
16939 The second way to customize your @code{text-mode-hook} is to set it
16940 yourself in your @file{.emacs} file using code that has nothing to do
16941 with the @code{custom-set-@dots{}} functions.
16944 When you do this, and later use @code{customize}, you will see a
16948 CHANGED outside Customize; operating on it here may be unreliable.
16952 This message is only a warning. If you click on the button to
16955 Save for Future Sessions
16959 Emacs will write a @code{custom-set-@dots{}} expression near the end
16960 of your @file{.emacs} file that will be evaluated after your
16961 hand-written expression. It will, therefore, overrule your
16962 hand-written expression. No harm will be done. When you do this,
16963 however, be careful to remember which expression is active; if you
16964 forget, you may confuse yourself.
16966 So long as you remember where the values are set, you will have no
16967 trouble. In any event, the values are always set in your
16968 initialization file, which is usually called @file{.emacs}.
16970 I myself use @code{customize} for hardly anything. Mostly, I write
16971 expressions myself.
16975 Incidentally, to be more complete concerning defines: @code{defsubst}
16976 defines an inline function. The syntax is just like that of
16977 @code{defun}. @code{defconst} defines a symbol as a constant. The
16978 intent is that neither programs nor users should ever change a value
16979 set by @code{defconst}. (You can change it; the value set is a
16980 variable; but please do not.)
16982 @node Beginning init File
16983 @section Beginning a @file{.emacs} File
16984 @cindex @file{.emacs} file, beginning of
16986 When you start Emacs, it loads your @file{.emacs} file unless you tell
16987 it not to by specifying @samp{-q} on the command line. (The
16988 @code{emacs -q} command gives you a plain, out-of-the-box Emacs.)
16990 A @file{.emacs} file contains Lisp expressions. Often, these are no
16991 more than expressions to set values; sometimes they are function
16994 @xref{Init File, , The Init File @file{~/.emacs}, emacs, The GNU Emacs
16995 Manual}, for a short description of initialization files.
16997 This chapter goes over some of the same ground, but is a walk among
16998 extracts from a complete, long-used @file{.emacs} file---my own.
17000 The first part of the file consists of comments: reminders to myself.
17001 By now, of course, I remember these things, but when I started, I did
17007 ;;;; Bob's .emacs file
17008 ; Robert J. Chassell
17009 ; 26 September 1985
17014 Look at that date! I started this file a long time ago. I have been
17015 adding to it ever since.
17019 ; Each section in this file is introduced by a
17020 ; line beginning with four semicolons; and each
17021 ; entry is introduced by a line beginning with
17022 ; three semicolons.
17027 This describes the usual conventions for comments in Emacs Lisp.
17028 Everything on a line that follows a semicolon is a comment. Two,
17029 three, and four semicolons are used as subsection and section markers.
17030 (@xref{Comments, ,, elisp, The GNU Emacs Lisp Reference Manual}, for
17031 more about comments.)
17036 ; Control-h is the help key;
17037 ; after typing control-h, type a letter to
17038 ; indicate the subject about which you want help.
17039 ; For an explanation of the help facility,
17040 ; type control-h two times in a row.
17045 Just remember: type @kbd{C-h} two times for help.
17049 ; To find out about any mode, type control-h m
17050 ; while in that mode. For example, to find out
17051 ; about mail mode, enter mail mode and then type
17057 `Mode help', as I call this, is very helpful. Usually, it tells you
17058 all you need to know.
17060 Of course, you don't need to include comments like these in your
17061 @file{.emacs} file. I included them in mine because I kept forgetting
17062 about Mode help or the conventions for comments---but I was able to
17063 remember to look here to remind myself.
17065 @node Text and Auto-fill
17066 @section Text and Auto Fill Mode
17068 Now we come to the part that `turns on' Text mode and
17073 ;;; Text mode and Auto Fill mode
17074 ;; The next two lines put Emacs into Text mode
17075 ;; and Auto Fill mode, and are for writers who
17076 ;; want to start writing prose rather than code.
17077 (setq-default major-mode 'text-mode)
17078 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17082 Here is the first part of this @file{.emacs} file that does something
17083 besides remind a forgetful human!
17085 The first of the two lines in parentheses tells Emacs to turn on Text
17086 mode when you find a file, @emph{unless} that file should go into some
17087 other mode, such as C mode.
17089 @cindex Per-buffer, local variables list
17090 @cindex Local variables list, per-buffer,
17091 @cindex Automatic mode selection
17092 @cindex Mode selection, automatic
17093 When Emacs reads a file, it looks at the extension to the file name,
17094 if any. (The extension is the part that comes after a @samp{.}.) If
17095 the file ends with a @samp{.c} or @samp{.h} extension then Emacs turns
17096 on C mode. Also, Emacs looks at first nonblank line of the file; if
17097 the line says @w{@samp{-*- C -*-}}, Emacs turns on C mode. Emacs
17098 possesses a list of extensions and specifications that it uses
17099 automatically. In addition, Emacs looks near the last page for a
17100 per-buffer, ``local variables list'', if any.
17103 @xref{Choosing Modes, , How Major Modes are Chosen, emacs, The GNU
17106 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17110 See sections ``How Major Modes are Chosen'' and ``Local Variables in
17111 Files'' in @cite{The GNU Emacs Manual}.
17114 Now, back to the @file{.emacs} file.
17117 Here is the line again; how does it work?
17119 @cindex Text Mode turned on
17121 (setq major-mode 'text-mode)
17125 This line is a short, but complete Emacs Lisp expression.
17127 We are already familiar with @code{setq}. It sets the following variable,
17128 @code{major-mode}, to the subsequent value, which is @code{text-mode}.
17129 The single quote mark before @code{text-mode} tells Emacs to deal directly
17130 with the @code{text-mode} symbol, not with whatever it might stand for.
17131 @xref{set & setq, , Setting the Value of a Variable},
17132 for a reminder of how @code{setq} works.
17133 The main point is that there is no difference between the procedure you
17134 use to set a value in your @file{.emacs} file and the procedure you use
17135 anywhere else in Emacs.
17138 Here is the next line:
17140 @cindex Auto Fill mode turned on
17143 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17147 In this line, the @code{add-hook} command adds
17148 @code{turn-on-auto-fill} to the variable.
17150 @code{turn-on-auto-fill} is the name of a program, that, you guessed
17151 it!, turns on Auto Fill mode.
17153 Every time Emacs turns on Text mode, Emacs runs the commands `hooked'
17154 onto Text mode. So every time Emacs turns on Text mode, Emacs also
17155 turns on Auto Fill mode.
17157 In brief, the first line causes Emacs to enter Text mode when you edit a
17158 file, unless the file name extension, a first non-blank line, or local
17159 variables to tell Emacs otherwise.
17161 Text mode among other actions, sets the syntax table to work
17162 conveniently for writers. In Text mode, Emacs considers an apostrophe
17163 as part of a word like a letter; but Emacs does not consider a period
17164 or a space as part of a word. Thus, @kbd{M-f} moves you over
17165 @samp{it's}. On the other hand, in C mode, @kbd{M-f} stops just after
17166 the @samp{t} of @samp{it's}.
17168 The second line causes Emacs to turn on Auto Fill mode when it turns
17169 on Text mode. In Auto Fill mode, Emacs automatically breaks a line
17170 that is too wide and brings the excessively wide part of the line down
17171 to the next line. Emacs breaks lines between words, not within them.
17173 When Auto Fill mode is turned off, lines continue to the right as you
17174 type them. Depending on how you set the value of
17175 @code{truncate-lines}, the words you type either disappear off the
17176 right side of the screen, or else are shown, in a rather ugly and
17177 unreadable manner, as a continuation line on the screen.
17180 In addition, in this part of my @file{.emacs} file, I tell the Emacs
17181 fill commands to insert two spaces after a colon:
17184 (setq colon-double-space t)
17188 @section Mail Aliases
17190 Here is a @code{setq} that `turns on' mail aliases, along with more
17196 ; To enter mail mode, type `C-x m'
17197 ; To enter RMAIL (for reading mail),
17199 (setq mail-aliases t)
17203 @cindex Mail aliases
17205 This @code{setq} command sets the value of the variable
17206 @code{mail-aliases} to @code{t}. Since @code{t} means true, the line
17207 says, in effect, ``Yes, use mail aliases.''
17209 Mail aliases are convenient short names for long email addresses or
17210 for lists of email addresses. The file where you keep your `aliases'
17211 is @file{~/.mailrc}. You write an alias like this:
17214 alias geo george@@foobar.wiz.edu
17218 When you write a message to George, address it to @samp{geo}; the
17219 mailer will automatically expand @samp{geo} to the full address.
17221 @node Indent Tabs Mode
17222 @section Indent Tabs Mode
17223 @cindex Tabs, preventing
17224 @findex indent-tabs-mode
17226 By default, Emacs inserts tabs in place of multiple spaces when it
17227 formats a region. (For example, you might indent many lines of text
17228 all at once with the @code{indent-region} command.) Tabs look fine on
17229 a terminal or with ordinary printing, but they produce badly indented
17230 output when you use @TeX{} or Texinfo since @TeX{} ignores tabs.
17233 The following turns off Indent Tabs mode:
17237 ;;; Prevent Extraneous Tabs
17238 (setq-default indent-tabs-mode nil)
17242 Note that this line uses @code{setq-default} rather than the
17243 @code{setq} command that we have seen before. The @code{setq-default}
17244 command sets values only in buffers that do not have their own local
17245 values for the variable.
17248 @xref{Just Spaces, , Tabs vs. Spaces, emacs, The GNU Emacs Manual}.
17250 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17254 See sections ``Tabs vs.@: Spaces'' and ``Local Variables in
17255 Files'' in @cite{The GNU Emacs Manual}.
17260 @section Some Keybindings
17262 Now for some personal keybindings:
17266 ;;; Compare windows
17267 (global-set-key "\C-cw" 'compare-windows)
17271 @findex compare-windows
17272 @code{compare-windows} is a nifty command that compares the text in
17273 your current window with text in the next window. It makes the
17274 comparison by starting at point in each window, moving over text in
17275 each window as far as they match. I use this command all the time.
17277 This also shows how to set a key globally, for all modes.
17279 @cindex Setting a key globally
17280 @cindex Global set key
17281 @cindex Key setting globally
17282 @findex global-set-key
17283 The command is @code{global-set-key}. It is followed by the
17284 keybinding. In a @file{.emacs} file, the keybinding is written as
17285 shown: @code{\C-c} stands for `control-c', which means `press the
17286 control key and the @key{c} key at the same time'. The @code{w} means
17287 `press the @key{w} key'. The keybinding is surrounded by double
17288 quotation marks. In documentation, you would write this as
17289 @w{@kbd{C-c w}}. (If you were binding a @key{META} key, such as
17290 @kbd{M-c}, rather than a @key{CTRL} key, you would write
17291 @w{@code{\M-c}} in your @file{.emacs} file. @xref{Init Rebinding, ,
17292 Rebinding Keys in Your Init File, emacs, The GNU Emacs Manual}, for
17295 The command invoked by the keys is @code{compare-windows}. Note that
17296 @code{compare-windows} is preceded by a single quote; otherwise, Emacs
17297 would first try to evaluate the symbol to determine its value.
17299 These three things, the double quotation marks, the backslash before
17300 the @samp{C}, and the single quote mark are necessary parts of
17301 keybinding that I tend to forget. Fortunately, I have come to
17302 remember that I should look at my existing @file{.emacs} file, and
17303 adapt what is there.
17305 As for the keybinding itself: @kbd{C-c w}. This combines the prefix
17306 key, @kbd{C-c}, with a single character, in this case, @kbd{w}. This
17307 set of keys, @kbd{C-c} followed by a single character, is strictly
17308 reserved for individuals' own use. (I call these `own' keys, since
17309 these are for my own use.) You should always be able to create such a
17310 keybinding for your own use without stomping on someone else's
17311 keybinding. If you ever write an extension to Emacs, please avoid
17312 taking any of these keys for public use. Create a key like @kbd{C-c
17313 C-w} instead. Otherwise, we will run out of `own' keys.
17316 Here is another keybinding, with a comment:
17320 ;;; Keybinding for `occur'
17321 ; I use occur a lot, so let's bind it to a key:
17322 (global-set-key "\C-co" 'occur)
17327 The @code{occur} command shows all the lines in the current buffer
17328 that contain a match for a regular expression. Matching lines are
17329 shown in a buffer called @file{*Occur*}. That buffer serves as a menu
17330 to jump to occurrences.
17332 @findex global-unset-key
17333 @cindex Unbinding key
17334 @cindex Key unbinding
17336 Here is how to unbind a key, so it does not
17342 (global-unset-key "\C-xf")
17346 There is a reason for this unbinding: I found I inadvertently typed
17347 @w{@kbd{C-x f}} when I meant to type @kbd{C-x C-f}. Rather than find a
17348 file, as I intended, I accidentally set the width for filled text,
17349 almost always to a width I did not want. Since I hardly ever reset my
17350 default width, I simply unbound the key.
17352 @findex list-buffers, @r{rebound}
17353 @findex buffer-menu, @r{bound to key}
17355 The following rebinds an existing key:
17359 ;;; Rebind `C-x C-b' for `buffer-menu'
17360 (global-set-key "\C-x\C-b" 'buffer-menu)
17364 By default, @kbd{C-x C-b} runs the
17365 @code{list-buffers} command. This command lists
17366 your buffers in @emph{another} window. Since I
17367 almost always want to do something in that
17368 window, I prefer the @code{buffer-menu}
17369 command, which not only lists the buffers,
17370 but moves point into that window.
17375 @cindex Rebinding keys
17377 Emacs uses @dfn{keymaps} to record which keys call which commands.
17378 When you use @code{global-set-key} to set the keybinding for a single
17379 command in all parts of Emacs, you are specifying the keybinding in
17380 @code{current-global-map}.
17382 Specific modes, such as C mode or Text mode, have their own keymaps;
17383 the mode-specific keymaps override the global map that is shared by
17386 The @code{global-set-key} function binds, or rebinds, the global
17387 keymap. For example, the following binds the key @kbd{C-x C-b} to the
17388 function @code{buffer-menu}:
17391 (global-set-key "\C-x\C-b" 'buffer-menu)
17394 Mode-specific keymaps are bound using the @code{define-key} function,
17395 which takes a specific keymap as an argument, as well as the key and
17396 the command. For example, my @file{.emacs} file contains the
17397 following expression to bind the @code{texinfo-insert-@@group} command
17398 to @kbd{C-c C-c g}:
17402 (define-key texinfo-mode-map "\C-c\C-cg" 'texinfo-insert-@@group)
17407 The @code{texinfo-insert-@@group} function itself is a little extension
17408 to Texinfo mode that inserts @samp{@@group} into a Texinfo file. I
17409 use this command all the time and prefer to type the three strokes
17410 @kbd{C-c C-c g} rather than the six strokes @kbd{@@ g r o u p}.
17411 (@samp{@@group} and its matching @samp{@@end group} are commands that
17412 keep all enclosed text together on one page; many multi-line examples
17413 in this book are surrounded by @samp{@@group @dots{} @@end group}.)
17416 Here is the @code{texinfo-insert-@@group} function definition:
17420 (defun texinfo-insert-@@group ()
17421 "Insert the string @@group in a Texinfo buffer."
17423 (beginning-of-line)
17424 (insert "@@group\n"))
17428 (Of course, I could have used Abbrev mode to save typing, rather than
17429 write a function to insert a word; but I prefer key strokes consistent
17430 with other Texinfo mode key bindings.)
17432 You will see numerous @code{define-key} expressions in
17433 @file{loaddefs.el} as well as in the various mode libraries, such as
17434 @file{cc-mode.el} and @file{lisp-mode.el}.
17436 @xref{Key Bindings, , Customizing Key Bindings, emacs, The GNU Emacs
17437 Manual}, and @ref{Keymaps, , Keymaps, elisp, The GNU Emacs Lisp
17438 Reference Manual}, for more information about keymaps.
17440 @node Loading Files
17441 @section Loading Files
17442 @cindex Loading files
17445 Many people in the GNU Emacs community have written extensions to
17446 Emacs. As time goes by, these extensions are often included in new
17447 releases. For example, the Calendar and Diary packages are now part
17448 of the standard GNU Emacs, as is Calc.
17450 You can use a @code{load} command to evaluate a complete file and
17451 thereby install all the functions and variables in the file into Emacs.
17454 @c (auto-compression-mode t)
17457 (load "~/emacs/slowsplit")
17460 This evaluates, i.e., loads, the @file{slowsplit.el} file or if it
17461 exists, the faster, byte compiled @file{slowsplit.elc} file from the
17462 @file{emacs} sub-directory of your home directory. The file contains
17463 the function @code{split-window-quietly}, which John Robinson wrote in
17466 The @code{split-window-quietly} function splits a window with the
17467 minimum of redisplay. I installed it in 1989 because it worked well
17468 with the slow 1200 baud terminals I was then using. Nowadays, I only
17469 occasionally come across such a slow connection, but I continue to use
17470 the function because I like the way it leaves the bottom half of a
17471 buffer in the lower of the new windows and the top half in the upper
17475 To replace the key binding for the default
17476 @code{split-window-vertically}, you must also unset that key and bind
17477 the keys to @code{split-window-quietly}, like this:
17481 (global-unset-key "\C-x2")
17482 (global-set-key "\C-x2" 'split-window-quietly)
17487 If you load many extensions, as I do, then instead of specifying the
17488 exact location of the extension file, as shown above, you can specify
17489 that directory as part of Emacs's @code{load-path}. Then, when Emacs
17490 loads a file, it will search that directory as well as its default
17491 list of directories. (The default list is specified in @file{paths.h}
17492 when Emacs is built.)
17495 The following command adds your @file{~/emacs} directory to the
17496 existing load path:
17500 ;;; Emacs Load Path
17501 (setq load-path (cons "~/emacs" load-path))
17505 Incidentally, @code{load-library} is an interactive interface to the
17506 @code{load} function. The complete function looks like this:
17508 @findex load-library
17511 (defun load-library (library)
17512 "Load the library named LIBRARY.
17513 This is an interface to the function `load'."
17515 (list (completing-read "Load library: "
17516 (apply-partially 'locate-file-completion-table
17518 (get-load-suffixes)))))
17523 The name of the function, @code{load-library}, comes from the use of
17524 `library' as a conventional synonym for `file'. The source for the
17525 @code{load-library} command is in the @file{files.el} library.
17527 Another interactive command that does a slightly different job is
17528 @code{load-file}. @xref{Lisp Libraries, , Libraries of Lisp Code for
17529 Emacs, emacs, The GNU Emacs Manual}, for information on the
17530 distinction between @code{load-library} and this command.
17533 @section Autoloading
17536 Instead of installing a function by loading the file that contains it,
17537 or by evaluating the function definition, you can make the function
17538 available but not actually install it until it is first called. This
17539 is called @dfn{autoloading}.
17541 When you execute an autoloaded function, Emacs automatically evaluates
17542 the file that contains the definition, and then calls the function.
17544 Emacs starts quicker with autoloaded functions, since their libraries
17545 are not loaded right away; but you need to wait a moment when you
17546 first use such a function, while its containing file is evaluated.
17548 Rarely used functions are frequently autoloaded. The
17549 @file{loaddefs.el} library contains hundreds of autoloaded functions,
17550 from @code{bookmark-set} to @code{wordstar-mode}. Of course, you may
17551 come to use a `rare' function frequently. When you do, you should
17552 load that function's file with a @code{load} expression in your
17553 @file{.emacs} file.
17555 In my @file{.emacs} file, I load 14 libraries that contain functions
17556 that would otherwise be autoloaded. (Actually, it would have been
17557 better to include these files in my `dumped' Emacs, but I forgot.
17558 @xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
17559 Reference Manual}, and the @file{INSTALL} file for more about
17562 You may also want to include autoloaded expressions in your @file{.emacs}
17563 file. @code{autoload} is a built-in function that takes up to five
17564 arguments, the final three of which are optional. The first argument
17565 is the name of the function to be autoloaded; the second is the name
17566 of the file to be loaded. The third argument is documentation for the
17567 function, and the fourth tells whether the function can be called
17568 interactively. The fifth argument tells what type of
17569 object---@code{autoload} can handle a keymap or macro as well as a
17570 function (the default is a function).
17573 Here is a typical example:
17577 (autoload 'html-helper-mode
17578 "html-helper-mode" "Edit HTML documents" t)
17583 (@code{html-helper-mode} is an older alternative to @code{html-mode},
17584 which is a standard part of the distribution.)
17587 This expression autoloads the @code{html-helper-mode} function. It
17588 takes it from the @file{html-helper-mode.el} file (or from the byte
17589 compiled version @file{html-helper-mode.elc}, if that exists.) The
17590 file must be located in a directory specified by @code{load-path}.
17591 The documentation says that this is a mode to help you edit documents
17592 written in the HyperText Markup Language. You can call this mode
17593 interactively by typing @kbd{M-x html-helper-mode}. (You need to
17594 duplicate the function's regular documentation in the autoload
17595 expression because the regular function is not yet loaded, so its
17596 documentation is not available.)
17598 @xref{Autoload, , Autoload, elisp, The GNU Emacs Lisp Reference
17599 Manual}, for more information.
17601 @node Simple Extension
17602 @section A Simple Extension: @code{line-to-top-of-window}
17603 @findex line-to-top-of-window
17604 @cindex Simple extension in @file{.emacs} file
17606 Here is a simple extension to Emacs that moves the line point is on to
17607 the top of the window. I use this all the time, to make text easier
17610 You can put the following code into a separate file and then load it
17611 from your @file{.emacs} file, or you can include it within your
17612 @file{.emacs} file.
17615 Here is the definition:
17619 ;;; Line to top of window;
17620 ;;; replace three keystroke sequence C-u 0 C-l
17621 (defun line-to-top-of-window ()
17622 "Move the line point is on to top of window."
17629 Now for the keybinding.
17631 Nowadays, function keys as well as mouse button events and
17632 non-@sc{ascii} characters are written within square brackets, without
17633 quotation marks. (In Emacs version 18 and before, you had to write
17634 different function key bindings for each different make of terminal.)
17636 I bind @code{line-to-top-of-window} to my @key{F6} function key like
17640 (global-set-key [f6] 'line-to-top-of-window)
17643 For more information, see @ref{Init Rebinding, , Rebinding Keys in
17644 Your Init File, emacs, The GNU Emacs Manual}.
17646 @cindex Conditional 'twixt two versions of Emacs
17647 @cindex Version of Emacs, choosing
17648 @cindex Emacs version, choosing
17649 If you run two versions of GNU Emacs, such as versions 22 and 23, and
17650 use one @file{.emacs} file, you can select which code to evaluate with
17651 the following conditional:
17656 ((= 22 emacs-major-version)
17657 ;; evaluate version 22 code
17659 ((= 23 emacs-major-version)
17660 ;; evaluate version 23 code
17665 For example, recent versions blink
17666 their cursors by default. I hate such blinking, as well as other
17667 features, so I placed the following in my @file{.emacs}
17668 file@footnote{When I start instances of Emacs that do not load my
17669 @file{.emacs} file or any site file, I also turn off blinking:
17672 emacs -q --no-site-file -eval '(blink-cursor-mode nil)'
17674 @exdent Or nowadays, using an even more sophisticated set of options,
17682 (when (>= emacs-major-version 21)
17683 (blink-cursor-mode 0)
17684 ;; Insert newline when you press `C-n' (next-line)
17685 ;; at the end of the buffer
17686 (setq next-line-add-newlines t)
17689 ;; Turn on image viewing
17690 (auto-image-file-mode t)
17693 ;; Turn on menu bar (this bar has text)
17694 ;; (Use numeric argument to turn on)
17698 ;; Turn off tool bar (this bar has icons)
17699 ;; (Use numeric argument to turn on)
17700 (tool-bar-mode nil)
17703 ;; Turn off tooltip mode for tool bar
17704 ;; (This mode causes icon explanations to pop up)
17705 ;; (Use numeric argument to turn on)
17707 ;; If tooltips turned on, make tips appear promptly
17708 (setq tooltip-delay 0.1) ; default is 0.7 second
17714 @section X11 Colors
17716 You can specify colors when you use Emacs with the MIT X Windowing
17719 I dislike the default colors and specify my own.
17722 Here are the expressions in my @file{.emacs}
17723 file that set values:
17727 ;; Set cursor color
17728 (set-cursor-color "white")
17731 (set-mouse-color "white")
17733 ;; Set foreground and background
17734 (set-foreground-color "white")
17735 (set-background-color "darkblue")
17739 ;;; Set highlighting colors for isearch and drag
17740 (set-face-foreground 'highlight "white")
17741 (set-face-background 'highlight "blue")
17745 (set-face-foreground 'region "cyan")
17746 (set-face-background 'region "blue")
17750 (set-face-foreground 'secondary-selection "skyblue")
17751 (set-face-background 'secondary-selection "darkblue")
17755 ;; Set calendar highlighting colors
17756 (setq calendar-load-hook
17758 (set-face-foreground 'diary-face "skyblue")
17759 (set-face-background 'holiday-face "slate blue")
17760 (set-face-foreground 'holiday-face "white")))
17764 The various shades of blue soothe my eye and prevent me from seeing
17765 the screen flicker.
17767 Alternatively, I could have set my specifications in various X
17768 initialization files. For example, I could set the foreground,
17769 background, cursor, and pointer (i.e., mouse) colors in my
17770 @file{~/.Xresources} file like this:
17774 Emacs*foreground: white
17775 Emacs*background: darkblue
17776 Emacs*cursorColor: white
17777 Emacs*pointerColor: white
17781 In any event, since it is not part of Emacs, I set the root color of
17782 my X window in my @file{~/.xinitrc} file, like this@footnote{I also
17783 run more modern window managers, such as Enlightenment, Gnome, or KDE;
17784 in those cases, I often specify an image rather than a plain color.}:
17787 xsetroot -solid Navy -fg white &
17791 @node Miscellaneous
17792 @section Miscellaneous Settings for a @file{.emacs} File
17795 Here are a few miscellaneous settings:
17800 Set the shape and color of the mouse cursor:
17804 ; Cursor shapes are defined in
17805 ; `/usr/include/X11/cursorfont.h';
17806 ; for example, the `target' cursor is number 128;
17807 ; the `top_left_arrow' cursor is number 132.
17811 (let ((mpointer (x-get-resource "*mpointer"
17812 "*emacs*mpointer")))
17813 ;; If you have not set your mouse pointer
17814 ;; then set it, otherwise leave as is:
17815 (if (eq mpointer nil)
17816 (setq mpointer "132")) ; top_left_arrow
17819 (setq x-pointer-shape (string-to-int mpointer))
17820 (set-mouse-color "white"))
17825 Or you can set the values of a variety of features in an alist, like
17831 default-frame-alist
17832 '((cursor-color . "white")
17833 (mouse-color . "white")
17834 (foreground-color . "white")
17835 (background-color . "DodgerBlue4")
17836 ;; (cursor-type . bar)
17837 (cursor-type . box)
17840 (tool-bar-lines . 0)
17841 (menu-bar-lines . 1)
17845 "-Misc-Fixed-Medium-R-Normal--20-200-75-75-C-100-ISO8859-1")
17851 Convert @kbd{@key{CTRL}-h} into @key{DEL} and @key{DEL}
17852 into @kbd{@key{CTRL}-h}.@*
17853 (Some older keyboards needed this, although I have not seen the
17858 ;; Translate `C-h' to <DEL>.
17859 ; (keyboard-translate ?\C-h ?\C-?)
17861 ;; Translate <DEL> to `C-h'.
17862 (keyboard-translate ?\C-? ?\C-h)
17866 @item Turn off a blinking cursor!
17870 (if (fboundp 'blink-cursor-mode)
17871 (blink-cursor-mode -1))
17876 or start GNU Emacs with the command @code{emacs -nbc}.
17879 @item When using `grep'@*
17880 @samp{-i}@w{ } Ignore case distinctions@*
17881 @samp{-n}@w{ } Prefix each line of output with line number@*
17882 @samp{-H}@w{ } Print the filename for each match.@*
17883 @samp{-e}@w{ } Protect patterns beginning with a hyphen character, @samp{-}
17886 (setq grep-command "grep -i -nH -e ")
17890 @c Evidently, no longer needed in GNU Emacs 22
17892 item Automatically uncompress compressed files when visiting them
17895 (load "uncompress")
17900 @item Find an existing buffer, even if it has a different name@*
17901 This avoids problems with symbolic links.
17904 (setq find-file-existing-other-name t)
17907 @item Set your language environment and default input method
17911 (set-language-environment "latin-1")
17912 ;; Remember you can enable or disable multilingual text input
17913 ;; with the @code{toggle-input-method'} (@kbd{C-\}) command
17914 (setq default-input-method "latin-1-prefix")
17918 If you want to write with Chinese `GB' characters, set this instead:
17922 (set-language-environment "Chinese-GB")
17923 (setq default-input-method "chinese-tonepy")
17928 @subsubheading Fixing Unpleasant Key Bindings
17929 @cindex Key bindings, fixing
17930 @cindex Bindings, key, fixing unpleasant
17932 Some systems bind keys unpleasantly. Sometimes, for example, the
17933 @key{CTRL} key appears in an awkward spot rather than at the far left
17936 Usually, when people fix these sorts of keybindings, they do not
17937 change their @file{~/.emacs} file. Instead, they bind the proper keys
17938 on their consoles with the @code{loadkeys} or @code{install-keymap}
17939 commands in their boot script and then include @code{xmodmap} commands
17940 in their @file{.xinitrc} or @file{.Xsession} file for X Windows.
17948 loadkeys /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz
17950 install-keymap emacs2
17956 For a @file{.xinitrc} or @file{.Xsession} file when the @key{Caps
17957 Lock} key is at the far left of the home row:
17961 # Bind the key labeled `Caps Lock' to `Control'
17962 # (Such a broken user interface suggests that keyboard manufacturers
17963 # think that computers are typewriters from 1885.)
17965 xmodmap -e "clear Lock"
17966 xmodmap -e "add Control = Caps_Lock"
17972 In a @file{.xinitrc} or @file{.Xsession} file, to convert an @key{ALT}
17973 key to a @key{META} key:
17977 # Some ill designed keyboards have a key labeled ALT and no Meta
17978 xmodmap -e "keysym Alt_L = Meta_L Alt_L"
17984 @section A Modified Mode Line
17985 @vindex mode-line-format
17986 @cindex Mode line format
17988 Finally, a feature I really like: a modified mode line.
17990 When I work over a network, I forget which machine I am using. Also,
17991 I tend to I lose track of where I am, and which line point is on.
17993 So I reset my mode line to look like this:
17996 -:-- foo.texi rattlesnake:/home/bob/ Line 1 (Texinfo Fill) Top
17999 I am visiting a file called @file{foo.texi}, on my machine
18000 @file{rattlesnake} in my @file{/home/bob} buffer. I am on line 1, in
18001 Texinfo mode, and am at the top of the buffer.
18004 My @file{.emacs} file has a section that looks like this:
18008 ;; Set a Mode Line that tells me which machine, which directory,
18009 ;; and which line I am on, plus the other customary information.
18010 (setq-default mode-line-format
18014 "mouse-1: select window, mouse-2: delete others ..."))
18015 mode-line-mule-info
18017 mode-line-frame-identification
18021 mode-line-buffer-identification
18024 (system-name) 0 (string-match "\\..+" (system-name))))
18029 "mouse-1: select window, mouse-2: delete others ..."))
18030 (line-number-mode " Line %l ")
18036 "mouse-1: select window, mouse-2: delete others ..."))
18037 (:eval (mode-line-mode-name))
18040 #("%n" 0 2 (help-echo "mouse-2: widen" local-map (keymap ...)))
18049 Here, I redefine the default mode line. Most of the parts are from
18050 the original; but I make a few changes. I set the @emph{default} mode
18051 line format so as to permit various modes, such as Info, to override
18054 Many elements in the list are self-explanatory:
18055 @code{mode-line-modified} is a variable that tells whether the buffer
18056 has been modified, @code{mode-name} tells the name of the mode, and so
18057 on. However, the format looks complicated because of two features we
18058 have not discussed.
18060 @cindex Properties, in mode line example
18061 The first string in the mode line is a dash or hyphen, @samp{-}. In
18062 the old days, it would have been specified simply as @code{"-"}. But
18063 nowadays, Emacs can add properties to a string, such as highlighting
18064 or, as in this case, a help feature. If you place your mouse cursor
18065 over the hyphen, some help information appears (By default, you must
18066 wait seven-tenths of a second before the information appears. You can
18067 change that timing by changing the value of @code{tooltip-delay}.)
18070 The new string format has a special syntax:
18073 #("-" 0 1 (help-echo "mouse-1: select window, ..."))
18077 The @code{#(} begins a list. The first element of the list is the
18078 string itself, just one @samp{-}. The second and third
18079 elements specify the range over which the fourth element applies. A
18080 range starts @emph{after} a character, so a zero means the range
18081 starts just before the first character; a 1 means that the range ends
18082 just after the first character. The third element is the property for
18083 the range. It consists of a property list, a
18084 property name, in this case, @samp{help-echo}, followed by a value, in this
18085 case, a string. The second, third, and fourth elements of this new
18086 string format can be repeated.
18088 @xref{Text Properties, , Text Properties, elisp, The GNU Emacs Lisp
18089 Reference Manual}, and see @ref{Mode Line Format, , Mode Line Format,
18090 elisp, The GNU Emacs Lisp Reference Manual}, for more information.
18092 @code{mode-line-buffer-identification}
18093 displays the current buffer name. It is a list
18094 beginning @code{(#("%12b" 0 4 @dots{}}.
18095 The @code{#(} begins the list.
18097 The @samp{"%12b"} displays the current buffer name, using the
18098 @code{buffer-name} function with which we are familiar; the `12'
18099 specifies the maximum number of characters that will be displayed.
18100 When a name has fewer characters, whitespace is added to fill out to
18101 this number. (Buffer names can and often should be longer than 12
18102 characters; this length works well in a typical 80 column wide
18105 @code{:eval} says to evaluate the following form and use the result as
18106 a string to display. In this case, the expression displays the first
18107 component of the full system name. The end of the first component is
18108 a @samp{.} (`period'), so I use the @code{string-match} function to
18109 tell me the length of the first component. The substring from the
18110 zeroth character to that length is the name of the machine.
18113 This is the expression:
18118 (system-name) 0 (string-match "\\..+" (system-name))))
18122 @samp{%[} and @samp{%]} cause a pair of square brackets
18123 to appear for each recursive editing level. @samp{%n} says `Narrow'
18124 when narrowing is in effect. @samp{%P} tells you the percentage of
18125 the buffer that is above the bottom of the window, or `Top', `Bottom',
18126 or `All'. (A lower case @samp{p} tell you the percentage above the
18127 @emph{top} of the window.) @samp{%-} inserts enough dashes to fill
18130 Remember, ``You don't have to like Emacs to like it''---your own
18131 Emacs can have different colors, different commands, and different
18132 keys than a default Emacs.
18134 On the other hand, if you want to bring up a plain `out of the box'
18135 Emacs, with no customization, type:
18142 This will start an Emacs that does @emph{not} load your
18143 @file{~/.emacs} initialization file. A plain, default Emacs. Nothing
18150 GNU Emacs has two debuggers, @code{debug} and @code{edebug}. The
18151 first is built into the internals of Emacs and is always with you;
18152 the second requires that you instrument a function before you can use it.
18154 Both debuggers are described extensively in @ref{Debugging, ,
18155 Debugging Lisp Programs, elisp, The GNU Emacs Lisp Reference Manual}.
18156 In this chapter, I will walk through a short example of each.
18159 * debug:: How to use the built-in debugger.
18160 * debug-on-entry:: Start debugging when you call a function.
18161 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
18162 * edebug:: How to use Edebug, a source level debugger.
18163 * Debugging Exercises::
18167 @section @code{debug}
18170 Suppose you have written a function definition that is intended to
18171 return the sum of the numbers 1 through a given number. (This is the
18172 @code{triangle} function discussed earlier. @xref{Decrementing
18173 Example, , Example with Decrementing Counter}, for a discussion.)
18174 @c xref{Decrementing Loop,, Loop with a Decrementing Counter}, for a discussion.)
18176 However, your function definition has a bug. You have mistyped
18177 @samp{1=} for @samp{1-}. Here is the broken definition:
18179 @findex triangle-bugged
18182 (defun triangle-bugged (number)
18183 "Return sum of numbers 1 through NUMBER inclusive."
18185 (while (> number 0)
18186 (setq total (+ total number))
18187 (setq number (1= number))) ; @r{Error here.}
18192 If you are reading this in Info, you can evaluate this definition in
18193 the normal fashion. You will see @code{triangle-bugged} appear in the
18197 Now evaluate the @code{triangle-bugged} function with an
18201 (triangle-bugged 4)
18205 In a recent GNU Emacs, you will create and enter a @file{*Backtrace*}
18211 ---------- Buffer: *Backtrace* ----------
18212 Debugger entered--Lisp error: (void-function 1=)
18214 (setq number (1= number))
18215 (while (> number 0) (setq total (+ total number))
18216 (setq number (1= number)))
18217 (let ((total 0)) (while (> number 0) (setq total ...)
18218 (setq number ...)) total)
18222 eval((triangle-bugged 4))
18223 eval-last-sexp-1(nil)
18224 eval-last-sexp(nil)
18225 call-interactively(eval-last-sexp)
18226 ---------- Buffer: *Backtrace* ----------
18231 (I have reformatted this example slightly; the debugger does not fold
18232 long lines. As usual, you can quit the debugger by typing @kbd{q} in
18233 the @file{*Backtrace*} buffer.)
18235 In practice, for a bug as simple as this, the `Lisp error' line will
18236 tell you what you need to know to correct the definition. The
18237 function @code{1=} is `void'.
18241 In GNU Emacs 20 and before, you will see:
18244 Symbol's function definition is void:@: 1=
18248 which has the same meaning as the @file{*Backtrace*} buffer line in
18252 However, suppose you are not quite certain what is going on?
18253 You can read the complete backtrace.
18255 In this case, you need to run a recent GNU Emacs, which automatically
18256 starts the debugger that puts you in the @file{*Backtrace*} buffer; or
18257 else, you need to start the debugger manually as described below.
18259 Read the @file{*Backtrace*} buffer from the bottom up; it tells you
18260 what Emacs did that led to the error. Emacs made an interactive call
18261 to @kbd{C-x C-e} (@code{eval-last-sexp}), which led to the evaluation
18262 of the @code{triangle-bugged} expression. Each line above tells you
18263 what the Lisp interpreter evaluated next.
18266 The third line from the top of the buffer is
18269 (setq number (1= number))
18273 Emacs tried to evaluate this expression; in order to do so, it tried
18274 to evaluate the inner expression shown on the second line from the
18283 This is where the error occurred; as the top line says:
18286 Debugger entered--Lisp error: (void-function 1=)
18290 You can correct the mistake, re-evaluate the function definition, and
18291 then run your test again.
18293 @node debug-on-entry
18294 @section @code{debug-on-entry}
18295 @findex debug-on-entry
18297 A recent GNU Emacs starts the debugger automatically when your
18298 function has an error.
18301 GNU Emacs version 20 and before did not; it simply
18302 presented you with an error message. You had to start the debugger
18306 Incidentally, you can start the debugger manually for all versions of
18307 Emacs; the advantage is that the debugger runs even if you do not have
18308 a bug in your code. Sometimes your code will be free of bugs!
18310 You can enter the debugger when you call the function by calling
18311 @code{debug-on-entry}.
18318 M-x debug-on-entry RET triangle-bugged RET
18323 Now, evaluate the following:
18326 (triangle-bugged 5)
18330 All versions of Emacs will create a @file{*Backtrace*} buffer and tell
18331 you that it is beginning to evaluate the @code{triangle-bugged}
18336 ---------- Buffer: *Backtrace* ----------
18337 Debugger entered--entering a function:
18338 * triangle-bugged(5)
18339 eval((triangle-bugged 5))
18342 eval-last-sexp-1(nil)
18343 eval-last-sexp(nil)
18344 call-interactively(eval-last-sexp)
18345 ---------- Buffer: *Backtrace* ----------
18349 In the @file{*Backtrace*} buffer, type @kbd{d}. Emacs will evaluate
18350 the first expression in @code{triangle-bugged}; the buffer will look
18355 ---------- Buffer: *Backtrace* ----------
18356 Debugger entered--beginning evaluation of function call form:
18357 * (let ((total 0)) (while (> number 0) (setq total ...)
18358 (setq number ...)) total)
18359 * triangle-bugged(5)
18360 eval((triangle-bugged 5))
18363 eval-last-sexp-1(nil)
18364 eval-last-sexp(nil)
18365 call-interactively(eval-last-sexp)
18366 ---------- Buffer: *Backtrace* ----------
18371 Now, type @kbd{d} again, eight times, slowly. Each time you type
18372 @kbd{d}, Emacs will evaluate another expression in the function
18376 Eventually, the buffer will look like this:
18380 ---------- Buffer: *Backtrace* ----------
18381 Debugger entered--beginning evaluation of function call form:
18382 * (setq number (1= number))
18383 * (while (> number 0) (setq total (+ total number))
18384 (setq number (1= number)))
18387 * (let ((total 0)) (while (> number 0) (setq total ...)
18388 (setq number ...)) total)
18389 * triangle-bugged(5)
18390 eval((triangle-bugged 5))
18393 eval-last-sexp-1(nil)
18394 eval-last-sexp(nil)
18395 call-interactively(eval-last-sexp)
18396 ---------- Buffer: *Backtrace* ----------
18402 Finally, after you type @kbd{d} two more times, Emacs will reach the
18403 error, and the top two lines of the @file{*Backtrace*} buffer will look
18408 ---------- Buffer: *Backtrace* ----------
18409 Debugger entered--Lisp error: (void-function 1=)
18412 ---------- Buffer: *Backtrace* ----------
18416 By typing @kbd{d}, you were able to step through the function.
18418 You can quit a @file{*Backtrace*} buffer by typing @kbd{q} in it; this
18419 quits the trace, but does not cancel @code{debug-on-entry}.
18421 @findex cancel-debug-on-entry
18422 To cancel the effect of @code{debug-on-entry}, call
18423 @code{cancel-debug-on-entry} and the name of the function, like this:
18426 M-x cancel-debug-on-entry RET triangle-bugged RET
18430 (If you are reading this in Info, cancel @code{debug-on-entry} now.)
18432 @node debug-on-quit
18433 @section @code{debug-on-quit} and @code{(debug)}
18435 In addition to setting @code{debug-on-error} or calling @code{debug-on-entry},
18436 there are two other ways to start @code{debug}.
18438 @findex debug-on-quit
18439 You can start @code{debug} whenever you type @kbd{C-g}
18440 (@code{keyboard-quit}) by setting the variable @code{debug-on-quit} to
18441 @code{t}. This is useful for debugging infinite loops.
18444 @cindex @code{(debug)} in code
18445 Or, you can insert a line that says @code{(debug)} into your code
18446 where you want the debugger to start, like this:
18450 (defun triangle-bugged (number)
18451 "Return sum of numbers 1 through NUMBER inclusive."
18453 (while (> number 0)
18454 (setq total (+ total number))
18455 (debug) ; @r{Start debugger.}
18456 (setq number (1= number))) ; @r{Error here.}
18461 The @code{debug} function is described in detail in @ref{Debugger, ,
18462 The Lisp Debugger, elisp, The GNU Emacs Lisp Reference Manual}.
18465 @section The @code{edebug} Source Level Debugger
18466 @cindex Source level debugger
18469 Edebug is a source level debugger. Edebug normally displays the
18470 source of the code you are debugging, with an arrow at the left that
18471 shows which line you are currently executing.
18473 You can walk through the execution of a function, line by line, or run
18474 quickly until reaching a @dfn{breakpoint} where execution stops.
18476 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18477 Lisp Reference Manual}.
18480 Here is a bugged function definition for @code{triangle-recursively}.
18481 @xref{Recursive triangle function, , Recursion in place of a counter},
18482 for a review of it.
18486 (defun triangle-recursively-bugged (number)
18487 "Return sum of numbers 1 through NUMBER inclusive.
18492 (triangle-recursively-bugged
18493 (1= number))))) ; @r{Error here.}
18498 Normally, you would install this definition by positioning your cursor
18499 after the function's closing parenthesis and typing @kbd{C-x C-e}
18500 (@code{eval-last-sexp}) or else by positioning your cursor within the
18501 definition and typing @kbd{C-M-x} (@code{eval-defun}). (By default,
18502 the @code{eval-defun} command works only in Emacs Lisp mode or in Lisp
18506 However, to prepare this function definition for Edebug, you must
18507 first @dfn{instrument} the code using a different command. You can do
18508 this by positioning your cursor within or just after the definition
18512 M-x edebug-defun RET
18516 This will cause Emacs to load Edebug automatically if it is not
18517 already loaded, and properly instrument the function.
18519 After instrumenting the function, place your cursor after the
18520 following expression and type @kbd{C-x C-e} (@code{eval-last-sexp}):
18523 (triangle-recursively-bugged 3)
18527 You will be jumped back to the source for
18528 @code{triangle-recursively-bugged} and the cursor positioned at the
18529 beginning of the @code{if} line of the function. Also, you will see
18530 an arrowhead at the left hand side of that line. The arrowhead marks
18531 the line where the function is executing. (In the following examples,
18532 we show the arrowhead with @samp{=>}; in a windowing system, you may
18533 see the arrowhead as a solid triangle in the window `fringe'.)
18536 =>@point{}(if (= number 1)
18541 In the example, the location of point is displayed with a star,
18542 @samp{@point{}} (in Info, it is displayed as @samp{-!-}).
18545 In the example, the location of point is displayed as @samp{@point{}}
18546 (in a printed book, it is displayed with a five pointed star).
18549 If you now press @key{SPC}, point will move to the next expression to
18550 be executed; the line will look like this:
18553 =>(if @point{}(= number 1)
18557 As you continue to press @key{SPC}, point will move from expression to
18558 expression. At the same time, whenever an expression returns a value,
18559 that value will be displayed in the echo area. For example, after you
18560 move point past @code{number}, you will see the following:
18563 Result: 3 (#o3, #x3, ?\C-c)
18567 This means the value of @code{number} is 3, which is octal three,
18568 hexadecimal three, and @sc{ascii} `control-c' (the third letter of the
18569 alphabet, in case you need to know this information).
18571 You can continue moving through the code until you reach the line with
18572 the error. Before evaluation, that line looks like this:
18575 => @point{}(1= number))))) ; @r{Error here.}
18580 When you press @key{SPC} once again, you will produce an error message
18584 Symbol's function definition is void:@: 1=
18590 Press @kbd{q} to quit Edebug.
18592 To remove instrumentation from a function definition, simply
18593 re-evaluate it with a command that does not instrument it.
18594 For example, you could place your cursor after the definition's
18595 closing parenthesis and type @kbd{C-x C-e}.
18597 Edebug does a great deal more than walk with you through a function.
18598 You can set it so it races through on its own, stopping only at an
18599 error or at specified stopping points; you can cause it to display the
18600 changing values of various expressions; you can find out how many
18601 times a function is called, and more.
18603 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18604 Lisp Reference Manual}.
18607 @node Debugging Exercises
18608 @section Debugging Exercises
18612 Install the @code{@value{COUNT-WORDS}} function and then cause it to
18613 enter the built-in debugger when you call it. Run the command on a
18614 region containing two words. You will need to press @kbd{d} a
18615 remarkable number of times. On your system, is a `hook' called after
18616 the command finishes? (For information on hooks, see @ref{Command
18617 Overview, , Command Loop Overview, elisp, The GNU Emacs Lisp Reference
18621 Copy @code{@value{COUNT-WORDS}} into the @file{*scratch*} buffer,
18622 instrument the function for Edebug, and walk through its execution.
18623 The function does not need to have a bug, although you can introduce
18624 one if you wish. If the function lacks a bug, the walk-through
18625 completes without problems.
18628 While running Edebug, type @kbd{?} to see a list of all the Edebug commands.
18629 (The @code{global-edebug-prefix} is usually @kbd{C-x X}, i.e.,
18630 @kbd{@key{CTRL}-x} followed by an upper case @kbd{X}; use this prefix
18631 for commands made outside of the Edebug debugging buffer.)
18634 In the Edebug debugging buffer, use the @kbd{p}
18635 (@code{edebug-bounce-point}) command to see where in the region the
18636 @code{@value{COUNT-WORDS}} is working.
18639 Move point to some spot further down the function and then type the
18640 @kbd{h} (@code{edebug-goto-here}) command to jump to that location.
18643 Use the @kbd{t} (@code{edebug-trace-mode}) command to cause Edebug to
18644 walk through the function on its own; use an upper case @kbd{T} for
18645 @code{edebug-Trace-fast-mode}.
18648 Set a breakpoint, then run Edebug in Trace mode until it reaches the
18653 @chapter Conclusion
18655 We have now reached the end of this Introduction. You have now
18656 learned enough about programming in Emacs Lisp to set values, to write
18657 simple @file{.emacs} files for yourself and your friends, and write
18658 simple customizations and extensions to Emacs.
18660 This is a place to stop. Or, if you wish, you can now go onward, and
18663 You have learned some of the basic nuts and bolts of programming. But
18664 only some. There are a great many more brackets and hinges that are
18665 easy to use that we have not touched.
18667 A path you can follow right now lies among the sources to GNU Emacs
18670 @cite{The GNU Emacs Lisp Reference Manual}.
18673 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
18674 Emacs Lisp Reference Manual}.
18677 The Emacs Lisp sources are an adventure. When you read the sources and
18678 come across a function or expression that is unfamiliar, you need to
18679 figure out or find out what it does.
18681 Go to the Reference Manual. It is a thorough, complete, and fairly
18682 easy-to-read description of Emacs Lisp. It is written not only for
18683 experts, but for people who know what you know. (The @cite{Reference
18684 Manual} comes with the standard GNU Emacs distribution. Like this
18685 introduction, it comes as a Texinfo source file, so you can read it
18686 on-line and as a typeset, printed book.)
18688 Go to the other on-line help that is part of GNU Emacs: the on-line
18689 documentation for all functions and variables, and @code{find-tag},
18690 the program that takes you to sources.
18692 Here is an example of how I explore the sources. Because of its name,
18693 @file{simple.el} is the file I looked at first, a long time ago. As
18694 it happens some of the functions in @file{simple.el} are complicated,
18695 or at least look complicated at first sight. The @code{open-line}
18696 function, for example, looks complicated.
18698 You may want to walk through this function slowly, as we did with the
18699 @code{forward-sentence} function. (@xref{forward-sentence, The
18700 @code{forward-sentence} function}.) Or you may want to skip that
18701 function and look at another, such as @code{split-line}. You don't
18702 need to read all the functions. According to
18703 @code{count-words-in-defun}, the @code{split-line} function contains
18704 102 words and symbols.
18706 Even though it is short, @code{split-line} contains expressions
18707 we have not studied: @code{skip-chars-forward}, @code{indent-to},
18708 @code{current-column} and @code{insert-and-inherit}.
18710 Consider the @code{skip-chars-forward} function. (It is part of the
18711 function definition for @code{back-to-indentation}, which is shown in
18712 @ref{Review, , Review}.)
18714 In GNU Emacs, you can find out more about @code{skip-chars-forward} by
18715 typing @kbd{C-h f} (@code{describe-function}) and the name of the
18716 function. This gives you the function documentation.
18718 You may be able to guess what is done by a well named function such as
18719 @code{indent-to}; or you can look it up, too. Incidentally, the
18720 @code{describe-function} function itself is in @file{help.el}; it is
18721 one of those long, but decipherable functions. You can look up
18722 @code{describe-function} using the @kbd{C-h f} command!
18724 In this instance, since the code is Lisp, the @file{*Help*} buffer
18725 contains the name of the library containing the function's source.
18726 You can put point over the name of the library and press the RET key,
18727 which in this situation is bound to @code{help-follow}, and be taken
18728 directly to the source, in the same way as @kbd{M-.}
18731 The definition for @code{describe-function} illustrates how to
18732 customize the @code{interactive} expression without using the standard
18733 character codes; and it shows how to create a temporary buffer.
18735 (The @code{indent-to} function is written in C rather than Emacs Lisp;
18736 it is a `built-in' function. @code{help-follow} takes you to its
18737 source as does @code{find-tag}, when properly set up.)
18739 You can look at a function's source using @code{find-tag}, which is
18740 bound to @kbd{M-.} Finally, you can find out what the Reference
18741 Manual has to say by visiting the manual in Info, and typing @kbd{i}
18742 (@code{Info-index}) and the name of the function, or by looking up the
18743 function in the index to a printed copy of the manual.
18745 Similarly, you can find out what is meant by
18746 @code{insert-and-inherit}.
18748 Other interesting source files include @file{paragraphs.el},
18749 @file{loaddefs.el}, and @file{loadup.el}. The @file{paragraphs.el}
18750 file includes short, easily understood functions as well as longer
18751 ones. The @file{loaddefs.el} file contains the many standard
18752 autoloads and many keymaps. I have never looked at it all; only at
18753 parts. @file{loadup.el} is the file that loads the standard parts of
18754 Emacs; it tells you a great deal about how Emacs is built.
18755 (@xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
18756 Reference Manual}, for more about building.)
18758 As I said, you have learned some nuts and bolts; however, and very
18759 importantly, we have hardly touched major aspects of programming; I
18760 have said nothing about how to sort information, except to use the
18761 predefined @code{sort} function; I have said nothing about how to store
18762 information, except to use variables and lists; I have said nothing
18763 about how to write programs that write programs. These are topics for
18764 another, and different kind of book, a different kind of learning.
18766 What you have done is learn enough for much practical work with GNU
18767 Emacs. What you have done is get started. This is the end of a
18770 @c ================ Appendix ================
18773 @appendix The @code{the-the} Function
18775 @cindex Duplicated words function
18776 @cindex Words, duplicated
18778 Sometimes when you you write text, you duplicate words---as with ``you
18779 you'' near the beginning of this sentence. I find that most
18780 frequently, I duplicate ``the''; hence, I call the function for
18781 detecting duplicated words, @code{the-the}.
18784 As a first step, you could use the following regular expression to
18785 search for duplicates:
18788 \\(\\w+[ \t\n]+\\)\\1
18792 This regexp matches one or more word-constituent characters followed
18793 by one or more spaces, tabs, or newlines. However, it does not detect
18794 duplicated words on different lines, since the ending of the first
18795 word, the end of the line, is different from the ending of the second
18796 word, a space. (For more information about regular expressions, see
18797 @ref{Regexp Search, , Regular Expression Searches}, as well as
18798 @ref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
18799 Manual}, and @ref{Regular Expressions, , Regular Expressions, elisp,
18800 The GNU Emacs Lisp Reference Manual}.)
18802 You might try searching just for duplicated word-constituent
18803 characters but that does not work since the pattern detects doubles
18804 such as the two occurrences of `th' in `with the'.
18806 Another possible regexp searches for word-constituent characters
18807 followed by non-word-constituent characters, reduplicated. Here,
18808 @w{@samp{\\w+}} matches one or more word-constituent characters and
18809 @w{@samp{\\W*}} matches zero or more non-word-constituent characters.
18812 \\(\\(\\w+\\)\\W*\\)\\1
18818 Here is the pattern that I use. It is not perfect, but good enough.
18819 @w{@samp{\\b}} matches the empty string, provided it is at the beginning
18820 or end of a word; @w{@samp{[^@@ \n\t]+}} matches one or more occurrences of
18821 any characters that are @emph{not} an @@-sign, space, newline, or tab.
18824 \\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b
18827 One can write more complicated expressions, but I found that this
18828 expression is good enough, so I use it.
18830 Here is the @code{the-the} function, as I include it in my
18831 @file{.emacs} file, along with a handy global key binding:
18836 "Search forward for for a duplicated word."
18838 (message "Searching for for duplicated words ...")
18842 ;; This regexp is not perfect
18843 ;; but is fairly good over all:
18844 (if (re-search-forward
18845 "\\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b" nil 'move)
18846 (message "Found duplicated word.")
18847 (message "End of buffer")))
18851 ;; Bind `the-the' to C-c \
18852 (global-set-key "\C-c\\" 'the-the)
18861 one two two three four five
18866 You can substitute the other regular expressions shown above in the
18867 function definition and try each of them on this list.
18870 @appendix Handling the Kill Ring
18871 @cindex Kill ring handling
18872 @cindex Handling the kill ring
18873 @cindex Ring, making a list like a
18875 The kill ring is a list that is transformed into a ring by the
18876 workings of the @code{current-kill} function. The @code{yank} and
18877 @code{yank-pop} commands use the @code{current-kill} function.
18879 This appendix describes the @code{current-kill} function as well as
18880 both the @code{yank} and the @code{yank-pop} commands, but first,
18881 consider the workings of the kill ring.
18884 * What the Kill Ring Does::
18886 * yank:: Paste a copy of a clipped element.
18887 * yank-pop:: Insert element pointed to.
18892 @node What the Kill Ring Does
18893 @unnumberedsec What the Kill Ring Does
18897 The kill ring has a default maximum length of sixty items; this number
18898 is too large for an explanation. Instead, set it to four. Please
18899 evaluate the following:
18903 (setq old-kill-ring-max kill-ring-max)
18904 (setq kill-ring-max 4)
18909 Then, please copy each line of the following indented example into the
18910 kill ring. You may kill each line with @kbd{C-k} or mark it and copy
18914 (In a read-only buffer, such as the @file{*info*} buffer, the kill
18915 command, @kbd{C-k} (@code{kill-line}), will not remove the text,
18916 merely copy it to the kill ring. However, your machine may beep at
18917 you. Alternatively, for silence, you may copy the region of each line
18918 with the @kbd{M-w} (@code{kill-ring-save}) command. You must mark
18919 each line for this command to succeed, but it does not matter at which
18920 end you put point or mark.)
18924 Please invoke the calls in order, so that five elements attempt to
18925 fill the kill ring:
18930 second piece of text
18932 fourth line of text
18939 Then find the value of @code{kill-ring} by evaluating
18951 ("fifth bit of text" "fourth line of text"
18952 "third line" "second piece of text")
18957 The first element, @samp{first some text}, was dropped.
18960 To return to the old value for the length of the kill ring, evaluate:
18963 (setq kill-ring-max old-kill-ring-max)
18967 @appendixsec The @code{current-kill} Function
18968 @findex current-kill
18970 The @code{current-kill} function changes the element in the kill ring
18971 to which @code{kill-ring-yank-pointer} points. (Also, the
18972 @code{kill-new} function sets @code{kill-ring-yank-pointer} to point
18973 to the latest element of the kill ring. The @code{kill-new}
18974 function is used directly or indirectly by @code{kill-append},
18975 @code{copy-region-as-kill}, @code{kill-ring-save}, @code{kill-line},
18976 and @code{kill-region}.)
18979 * Code for current-kill::
18980 * Understanding current-kill::
18984 @node Code for current-kill
18985 @unnumberedsubsec The code for @code{current-kill}
18990 The @code{current-kill} function is used by @code{yank} and by
18991 @code{yank-pop}. Here is the code for @code{current-kill}:
18995 (defun current-kill (n &optional do-not-move)
18996 "Rotate the yanking point by N places, and then return that kill.
18997 If N is zero, `interprogram-paste-function' is set, and calling it
18998 returns a string, then that string is added to the front of the
18999 kill ring and returned as the latest kill.
19002 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
19003 yanking point; just return the Nth kill forward."
19004 (let ((interprogram-paste (and (= n 0)
19005 interprogram-paste-function
19006 (funcall interprogram-paste-function))))
19009 (if interprogram-paste
19011 ;; Disable the interprogram cut function when we add the new
19012 ;; text to the kill ring, so Emacs doesn't try to own the
19013 ;; selection, with identical text.
19014 (let ((interprogram-cut-function nil))
19015 (kill-new interprogram-paste))
19016 interprogram-paste)
19019 (or kill-ring (error "Kill ring is empty"))
19020 (let ((ARGth-kill-element
19021 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19022 (length kill-ring))
19025 (setq kill-ring-yank-pointer ARGth-kill-element))
19026 (car ARGth-kill-element)))))
19030 Remember also that the @code{kill-new} function sets
19031 @code{kill-ring-yank-pointer} to the latest element of the kill
19032 ring, which means that all the functions that call it set the value
19033 indirectly: @code{kill-append}, @code{copy-region-as-kill},
19034 @code{kill-ring-save}, @code{kill-line}, and @code{kill-region}.
19037 Here is the line in @code{kill-new}, which is explained in
19038 @ref{kill-new function, , The @code{kill-new} function}.
19041 (setq kill-ring-yank-pointer kill-ring)
19045 @node Understanding current-kill
19046 @unnumberedsubsec @code{current-kill} in Outline
19049 The @code{current-kill} function looks complex, but as usual, it can
19050 be understood by taking it apart piece by piece. First look at it in
19055 (defun current-kill (n &optional do-not-move)
19056 "Rotate the yanking point by N places, and then return that kill."
19062 This function takes two arguments, one of which is optional. It has a
19063 documentation string. It is @emph{not} interactive.
19066 * Body of current-kill::
19067 * Digression concerning error:: How to mislead humans, but not computers.
19068 * Determining the Element::
19072 @node Body of current-kill
19073 @unnumberedsubsubsec The Body of @code{current-kill}
19076 The body of the function definition is a @code{let} expression, which
19077 itself has a body as well as a @var{varlist}.
19079 The @code{let} expression declares a variable that will be only usable
19080 within the bounds of this function. This variable is called
19081 @code{interprogram-paste} and is for copying to another program. It
19082 is not for copying within this instance of GNU Emacs. Most window
19083 systems provide a facility for interprogram pasting. Sadly, that
19084 facility usually provides only for the last element. Most windowing
19085 systems have not adopted a ring of many possibilities, even though
19086 Emacs has provided it for decades.
19088 The @code{if} expression has two parts, one if there exists
19089 @code{interprogram-paste} and one if not.
19092 Let us consider the `if not' or else-part of the @code{current-kill}
19093 function. (The then-part uses the @code{kill-new} function, which
19094 we have already described. @xref{kill-new function, , The
19095 @code{kill-new} function}.)
19099 (or kill-ring (error "Kill ring is empty"))
19100 (let ((ARGth-kill-element
19101 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19102 (length kill-ring))
19105 (setq kill-ring-yank-pointer ARGth-kill-element))
19106 (car ARGth-kill-element))
19111 The code first checks whether the kill ring has content; otherwise it
19115 Note that the @code{or} expression is very similar to testing length
19122 (if (zerop (length kill-ring)) ; @r{if-part}
19123 (error "Kill ring is empty")) ; @r{then-part}
19129 If there is not anything in the kill ring, its length must be zero and
19130 an error message sent to the user: @samp{Kill ring is empty}. The
19131 @code{current-kill} function uses an @code{or} expression which is
19132 simpler. But an @code{if} expression reminds us what goes on.
19134 This @code{if} expression uses the function @code{zerop} which returns
19135 true if the value it is testing is zero. When @code{zerop} tests
19136 true, the then-part of the @code{if} is evaluated. The then-part is a
19137 list starting with the function @code{error}, which is a function that
19138 is similar to the @code{message} function
19139 (@pxref{message, , The @code{message} Function}) in that
19140 it prints a one-line message in the echo area. However, in addition
19141 to printing a message, @code{error} also stops evaluation of the
19142 function within which it is embedded. This means that the rest of the
19143 function will not be evaluated if the length of the kill ring is zero.
19145 Then the @code{current-kill} function selects the element to return.
19146 The selection depends on the number of places that @code{current-kill}
19147 rotates and on where @code{kill-ring-yank-pointer} points.
19149 Next, either the optional @code{do-not-move} argument is true or the
19150 current value of @code{kill-ring-yank-pointer} is set to point to the
19151 list. Finally, another expression returns the first element of the
19152 list even if the @code{do-not-move} argument is true.
19155 @node Digression concerning error
19156 @unnumberedsubsubsec Digression about the word `error'
19159 In my opinion, it is slightly misleading, at least to humans, to use
19160 the term `error' as the name of the @code{error} function. A better
19161 term would be `cancel'. Strictly speaking, of course, you cannot
19162 point to, much less rotate a pointer to a list that has no length, so
19163 from the point of view of the computer, the word `error' is correct.
19164 But a human expects to attempt this sort of thing, if only to find out
19165 whether the kill ring is full or empty. This is an act of
19168 From the human point of view, the act of exploration and discovery is
19169 not necessarily an error, and therefore should not be labeled as one,
19170 even in the bowels of a computer. As it is, the code in Emacs implies
19171 that a human who is acting virtuously, by exploring his or her
19172 environment, is making an error. This is bad. Even though the computer
19173 takes the same steps as it does when there is an `error', a term such as
19174 `cancel' would have a clearer connotation.
19177 @node Determining the Element
19178 @unnumberedsubsubsec Determining the Element
19181 Among other actions, the else-part of the @code{if} expression sets
19182 the value of @code{kill-ring-yank-pointer} to
19183 @code{ARGth-kill-element} when the kill ring has something in it and
19184 the value of @code{do-not-move} is @code{nil}.
19187 The code looks like this:
19191 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19192 (length kill-ring))
19197 This needs some examination. Unless it is not supposed to move the
19198 pointer, the @code{current-kill} function changes where
19199 @code{kill-ring-yank-pointer} points.
19201 @w{@code{(setq kill-ring-yank-pointer ARGth-kill-element))}}
19202 expression does. Also, clearly, @code{ARGth-kill-element} is being
19203 set to be equal to some @sc{cdr} of the kill ring, using the
19204 @code{nthcdr} function that is described in an earlier section.
19205 (@xref{copy-region-as-kill}.) How does it do this?
19207 As we have seen before (@pxref{nthcdr}), the @code{nthcdr} function
19208 works by repeatedly taking the @sc{cdr} of a list---it takes the
19209 @sc{cdr} of the @sc{cdr} of the @sc{cdr} @dots{}
19212 The two following expressions produce the same result:
19216 (setq kill-ring-yank-pointer (cdr kill-ring))
19218 (setq kill-ring-yank-pointer (nthcdr 1 kill-ring))
19222 However, the @code{nthcdr} expression is more complicated. It uses
19223 the @code{mod} function to determine which @sc{cdr} to select.
19225 (You will remember to look at inner functions first; indeed, we will
19226 have to go inside the @code{mod}.)
19228 The @code{mod} function returns the value of its first argument modulo
19229 the second; that is to say, it returns the remainder after dividing
19230 the first argument by the second. The value returned has the same
19231 sign as the second argument.
19239 @result{} 0 ;; @r{because there is no remainder}
19246 In this case, the first argument is often smaller than the second.
19258 We can guess what the @code{-} function does. It is like @code{+} but
19259 subtracts instead of adds; the @code{-} function subtracts its second
19260 argument from its first. Also, we already know what the @code{length}
19261 function does (@pxref{length}). It returns the length of a list.
19263 And @code{n} is the name of the required argument to the
19264 @code{current-kill} function.
19267 So when the first argument to @code{nthcdr} is zero, the @code{nthcdr}
19268 expression returns the whole list, as you can see by evaluating the
19273 ;; kill-ring-yank-pointer @r{and} kill-ring @r{have a length of four}
19274 ;; @r{and} (mod (- 0 4) 4) @result{} 0
19275 (nthcdr (mod (- 0 4) 4)
19276 '("fourth line of text"
19278 "second piece of text"
19279 "first some text"))
19284 When the first argument to the @code{current-kill} function is one,
19285 the @code{nthcdr} expression returns the list without its first
19290 (nthcdr (mod (- 1 4) 4)
19291 '("fourth line of text"
19293 "second piece of text"
19294 "first some text"))
19298 @cindex @samp{global variable} defined
19299 @cindex @samp{variable, global}, defined
19300 Incidentally, both @code{kill-ring} and @code{kill-ring-yank-pointer}
19301 are @dfn{global variables}. That means that any expression in Emacs
19302 Lisp can access them. They are not like the local variables set by
19303 @code{let} or like the symbols in an argument list.
19304 Local variables can only be accessed
19305 within the @code{let} that defines them or the function that specifies
19306 them in an argument list (and within expressions called by them).
19309 @c texi2dvi fails when the name of the section is within ifnottex ...
19310 (@xref{Prevent confusion, , @code{let} Prevents Confusion}, and
19311 @ref{defun, , The @code{defun} Macro}.)
19315 @appendixsec @code{yank}
19318 After learning about @code{current-kill}, the code for the
19319 @code{yank} function is almost easy.
19321 The @code{yank} function does not use the
19322 @code{kill-ring-yank-pointer} variable directly. It calls
19323 @code{insert-for-yank} which calls @code{current-kill} which sets the
19324 @code{kill-ring-yank-pointer} variable.
19327 The code looks like this:
19332 (defun yank (&optional arg)
19333 "Reinsert (\"paste\") the last stretch of killed text.
19334 More precisely, reinsert the stretch of killed text most recently
19335 killed OR yanked. Put point at end, and set mark at beginning.
19336 With just \\[universal-argument] as argument, same but put point at
19337 beginning (and mark at end). With argument N, reinsert the Nth most
19338 recently killed stretch of killed text.
19340 When this command inserts killed text into the buffer, it honors
19341 `yank-excluded-properties' and `yank-handler' as described in the
19342 doc string for `insert-for-yank-1', which see.
19344 See also the command \\[yank-pop]."
19348 (setq yank-window-start (window-start))
19349 ;; If we don't get all the way thru, make last-command indicate that
19350 ;; for the following command.
19351 (setq this-command t)
19352 (push-mark (point))
19355 (insert-for-yank (current-kill (cond
19360 ;; This is like exchange-point-and-mark,
19361 ;; but doesn't activate the mark.
19362 ;; It is cleaner to avoid activation, even though the command
19363 ;; loop would deactivate the mark because we inserted text.
19364 (goto-char (prog1 (mark t)
19365 (set-marker (mark-marker) (point) (current-buffer)))))
19368 ;; If we do get all the way thru, make this-command indicate that.
19369 (if (eq this-command t)
19370 (setq this-command 'yank))
19375 The key expression is @code{insert-for-yank}, which inserts the string
19376 returned by @code{current-kill}, but removes some text properties from
19379 However, before getting to that expression, the function sets the value
19380 of @code{yank-window-start} to the position returned by the
19381 @code{(window-start)} expression, the position at which the display
19382 currently starts. The @code{yank} function also sets
19383 @code{this-command} and pushes the mark.
19385 After it yanks the appropriate element, if the optional argument is a
19386 @sc{cons} rather than a number or nothing, it puts point at beginning
19387 of the yanked text and mark at its end.
19389 (The @code{prog1} function is like @code{progn} but returns the value
19390 of its first argument rather than the value of its last argument. Its
19391 first argument is forced to return the buffer's mark as an integer.
19392 You can see the documentation for these functions by placing point
19393 over them in this buffer and then typing @kbd{C-h f}
19394 (@code{describe-function}) followed by a @kbd{RET}; the default is the
19397 The last part of the function tells what to do when it succeeds.
19400 @appendixsec @code{yank-pop}
19403 After understanding @code{yank} and @code{current-kill}, you know how
19404 to approach the @code{yank-pop} function. Leaving out the
19405 documentation to save space, it looks like this:
19410 (defun yank-pop (&optional arg)
19413 (if (not (eq last-command 'yank))
19414 (error "Previous command was not a yank"))
19417 (setq this-command 'yank)
19418 (unless arg (setq arg 1))
19419 (let ((inhibit-read-only t)
19420 (before (< (point) (mark t))))
19424 (funcall (or yank-undo-function 'delete-region) (point) (mark t))
19425 (funcall (or yank-undo-function 'delete-region) (mark t) (point)))
19426 (setq yank-undo-function nil)
19429 (set-marker (mark-marker) (point) (current-buffer))
19430 (insert-for-yank (current-kill arg))
19431 ;; Set the window start back where it was in the yank command,
19433 (set-window-start (selected-window) yank-window-start t)
19437 ;; This is like exchange-point-and-mark,
19438 ;; but doesn't activate the mark.
19439 ;; It is cleaner to avoid activation, even though the command
19440 ;; loop would deactivate the mark because we inserted text.
19441 (goto-char (prog1 (mark t)
19442 (set-marker (mark-marker)
19444 (current-buffer))))))
19449 The function is interactive with a small @samp{p} so the prefix
19450 argument is processed and passed to the function. The command can
19451 only be used after a previous yank; otherwise an error message is
19452 sent. This check uses the variable @code{last-command} which is set
19453 by @code{yank} and is discussed elsewhere.
19454 (@xref{copy-region-as-kill}.)
19456 The @code{let} clause sets the variable @code{before} to true or false
19457 depending whether point is before or after mark and then the region
19458 between point and mark is deleted. This is the region that was just
19459 inserted by the previous yank and it is this text that will be
19462 @code{funcall} calls its first argument as a function, passing
19463 remaining arguments to it. The first argument is whatever the
19464 @code{or} expression returns. The two remaining arguments are the
19465 positions of point and mark set by the preceding @code{yank} command.
19467 There is more, but that is the hardest part.
19470 @appendixsec The @file{ring.el} File
19471 @cindex @file{ring.el} file
19473 Interestingly, GNU Emacs posses a file called @file{ring.el} that
19474 provides many of the features we just discussed. But functions such
19475 as @code{kill-ring-yank-pointer} do not use this library, possibly
19476 because they were written earlier.
19479 @appendix A Graph with Labeled Axes
19481 Printed axes help you understand a graph. They convey scale. In an
19482 earlier chapter (@pxref{Readying a Graph, , Readying a Graph}), we
19483 wrote the code to print the body of a graph. Here we write the code
19484 for printing and labeling vertical and horizontal axes, along with the
19488 * Labeled Example::
19489 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
19490 * print-Y-axis:: Print a label for the vertical axis.
19491 * print-X-axis:: Print a horizontal label.
19492 * Print Whole Graph:: The function to print a complete graph.
19496 @node Labeled Example
19497 @unnumberedsec Labeled Example Graph
19500 Since insertions fill a buffer to the right and below point, the new
19501 graph printing function should first print the Y or vertical axis,
19502 then the body of the graph, and finally the X or horizontal axis.
19503 This sequence lays out for us the contents of the function:
19513 Print body of graph.
19520 Here is an example of how a finished graph should look:
19533 1 - ****************
19540 In this graph, both the vertical and the horizontal axes are labeled
19541 with numbers. However, in some graphs, the horizontal axis is time
19542 and would be better labeled with months, like this:
19556 Indeed, with a little thought, we can easily come up with a variety of
19557 vertical and horizontal labeling schemes. Our task could become
19558 complicated. But complications breed confusion. Rather than permit
19559 this, it is better choose a simple labeling scheme for our first
19560 effort, and to modify or replace it later.
19563 These considerations suggest the following outline for the
19564 @code{print-graph} function:
19568 (defun print-graph (numbers-list)
19569 "@var{documentation}@dots{}"
19570 (let ((height @dots{}
19574 (print-Y-axis height @dots{} )
19575 (graph-body-print numbers-list)
19576 (print-X-axis @dots{} )))
19580 We can work on each part of the @code{print-graph} function definition
19583 @node print-graph Varlist
19584 @appendixsec The @code{print-graph} Varlist
19585 @cindex @code{print-graph} varlist
19587 In writing the @code{print-graph} function, the first task is to write
19588 the varlist in the @code{let} expression. (We will leave aside for the
19589 moment any thoughts about making the function interactive or about the
19590 contents of its documentation string.)
19592 The varlist should set several values. Clearly, the top of the label
19593 for the vertical axis must be at least the height of the graph, which
19594 means that we must obtain this information here. Note that the
19595 @code{print-graph-body} function also requires this information. There
19596 is no reason to calculate the height of the graph in two different
19597 places, so we should change @code{print-graph-body} from the way we
19598 defined it earlier to take advantage of the calculation.
19600 Similarly, both the function for printing the X axis labels and the
19601 @code{print-graph-body} function need to learn the value of the width of
19602 each symbol. We can perform the calculation here and change the
19603 definition for @code{print-graph-body} from the way we defined it in the
19606 The length of the label for the horizontal axis must be at least as long
19607 as the graph. However, this information is used only in the function
19608 that prints the horizontal axis, so it does not need to be calculated here.
19610 These thoughts lead us directly to the following form for the varlist
19611 in the @code{let} for @code{print-graph}:
19615 (let ((height (apply 'max numbers-list)) ; @r{First version.}
19616 (symbol-width (length graph-blank)))
19621 As we shall see, this expression is not quite right.
19625 @appendixsec The @code{print-Y-axis} Function
19626 @cindex Axis, print vertical
19627 @cindex Y axis printing
19628 @cindex Vertical axis printing
19629 @cindex Print vertical axis
19631 The job of the @code{print-Y-axis} function is to print a label for
19632 the vertical axis that looks like this:
19650 The function should be passed the height of the graph, and then should
19651 construct and insert the appropriate numbers and marks.
19654 * print-Y-axis in Detail::
19655 * Height of label:: What height for the Y axis?
19656 * Compute a Remainder:: How to compute the remainder of a division.
19657 * Y Axis Element:: Construct a line for the Y axis.
19658 * Y-axis-column:: Generate a list of Y axis labels.
19659 * print-Y-axis Penultimate:: A not quite final version.
19663 @node print-Y-axis in Detail
19664 @unnumberedsubsec The @code{print-Y-axis} Function in Detail
19667 It is easy enough to see in the figure what the Y axis label should
19668 look like; but to say in words, and then to write a function
19669 definition to do the job is another matter. It is not quite true to
19670 say that we want a number and a tic every five lines: there are only
19671 three lines between the @samp{1} and the @samp{5} (lines 2, 3, and 4),
19672 but four lines between the @samp{5} and the @samp{10} (lines 6, 7, 8,
19673 and 9). It is better to say that we want a number and a tic mark on
19674 the base line (number 1) and then that we want a number and a tic on
19675 the fifth line from the bottom and on every line that is a multiple of
19679 @node Height of label
19680 @unnumberedsubsec What height should the label be?
19683 The next issue is what height the label should be? Suppose the maximum
19684 height of tallest column of the graph is seven. Should the highest
19685 label on the Y axis be @samp{5 -}, and should the graph stick up above
19686 the label? Or should the highest label be @samp{7 -}, and mark the peak
19687 of the graph? Or should the highest label be @code{10 -}, which is a
19688 multiple of five, and be higher than the topmost value of the graph?
19690 The latter form is preferred. Most graphs are drawn within rectangles
19691 whose sides are an integral number of steps long---5, 10, 15, and so
19692 on for a step distance of five. But as soon as we decide to use a
19693 step height for the vertical axis, we discover that the simple
19694 expression in the varlist for computing the height is wrong. The
19695 expression is @code{(apply 'max numbers-list)}. This returns the
19696 precise height, not the maximum height plus whatever is necessary to
19697 round up to the nearest multiple of five. A more complex expression
19700 As usual in cases like this, a complex problem becomes simpler if it is
19701 divided into several smaller problems.
19703 First, consider the case when the highest value of the graph is an
19704 integral multiple of five---when it is 5, 10, 15, or some higher
19705 multiple of five. We can use this value as the Y axis height.
19707 A fairly simply way to determine whether a number is a multiple of
19708 five is to divide it by five and see if the division results in a
19709 remainder. If there is no remainder, the number is a multiple of
19710 five. Thus, seven divided by five has a remainder of two, and seven
19711 is not an integral multiple of five. Put in slightly different
19712 language, more reminiscent of the classroom, five goes into seven
19713 once, with a remainder of two. However, five goes into ten twice,
19714 with no remainder: ten is an integral multiple of five.
19716 @node Compute a Remainder
19717 @appendixsubsec Side Trip: Compute a Remainder
19719 @findex % @r{(remainder function)}
19720 @cindex Remainder function, @code{%}
19721 In Lisp, the function for computing a remainder is @code{%}. The
19722 function returns the remainder of its first argument divided by its
19723 second argument. As it happens, @code{%} is a function in Emacs Lisp
19724 that you cannot discover using @code{apropos}: you find nothing if you
19725 type @kbd{M-x apropos @key{RET} remainder @key{RET}}. The only way to
19726 learn of the existence of @code{%} is to read about it in a book such
19727 as this or in the Emacs Lisp sources.
19729 You can try the @code{%} function by evaluating the following two
19741 The first expression returns 2 and the second expression returns 0.
19743 To test whether the returned value is zero or some other number, we
19744 can use the @code{zerop} function. This function returns @code{t} if
19745 its argument, which must be a number, is zero.
19757 Thus, the following expression will return @code{t} if the height
19758 of the graph is evenly divisible by five:
19761 (zerop (% height 5))
19765 (The value of @code{height}, of course, can be found from @code{(apply
19766 'max numbers-list)}.)
19768 On the other hand, if the value of @code{height} is not a multiple of
19769 five, we want to reset the value to the next higher multiple of five.
19770 This is straightforward arithmetic using functions with which we are
19771 already familiar. First, we divide the value of @code{height} by five
19772 to determine how many times five goes into the number. Thus, five
19773 goes into twelve twice. If we add one to this quotient and multiply by
19774 five, we will obtain the value of the next multiple of five that is
19775 larger than the height. Five goes into twelve twice. Add one to two,
19776 and multiply by five; the result is fifteen, which is the next multiple
19777 of five that is higher than twelve. The Lisp expression for this is:
19780 (* (1+ (/ height 5)) 5)
19784 For example, if you evaluate the following, the result is 15:
19787 (* (1+ (/ 12 5)) 5)
19790 All through this discussion, we have been using `five' as the value
19791 for spacing labels on the Y axis; but we may want to use some other
19792 value. For generality, we should replace `five' with a variable to
19793 which we can assign a value. The best name I can think of for this
19794 variable is @code{Y-axis-label-spacing}.
19797 Using this term, and an @code{if} expression, we produce the
19802 (if (zerop (% height Y-axis-label-spacing))
19805 (* (1+ (/ height Y-axis-label-spacing))
19806 Y-axis-label-spacing))
19811 This expression returns the value of @code{height} itself if the height
19812 is an even multiple of the value of the @code{Y-axis-label-spacing} or
19813 else it computes and returns a value of @code{height} that is equal to
19814 the next higher multiple of the value of the @code{Y-axis-label-spacing}.
19816 We can now include this expression in the @code{let} expression of the
19817 @code{print-graph} function (after first setting the value of
19818 @code{Y-axis-label-spacing}):
19819 @vindex Y-axis-label-spacing
19823 (defvar Y-axis-label-spacing 5
19824 "Number of lines from one Y axis label to next.")
19829 (let* ((height (apply 'max numbers-list))
19830 (height-of-top-line
19831 (if (zerop (% height Y-axis-label-spacing))
19836 (* (1+ (/ height Y-axis-label-spacing))
19837 Y-axis-label-spacing)))
19838 (symbol-width (length graph-blank))))
19844 (Note use of the @code{let*} function: the initial value of height is
19845 computed once by the @code{(apply 'max numbers-list)} expression and
19846 then the resulting value of @code{height} is used to compute its
19847 final value. @xref{fwd-para let, , The @code{let*} expression}, for
19848 more about @code{let*}.)
19850 @node Y Axis Element
19851 @appendixsubsec Construct a Y Axis Element
19853 When we print the vertical axis, we want to insert strings such as
19854 @w{@samp{5 -}} and @w{@samp{10 - }} every five lines.
19855 Moreover, we want the numbers and dashes to line up, so shorter
19856 numbers must be padded with leading spaces. If some of the strings
19857 use two digit numbers, the strings with single digit numbers must
19858 include a leading blank space before the number.
19860 @findex number-to-string
19861 To figure out the length of the number, the @code{length} function is
19862 used. But the @code{length} function works only with a string, not with
19863 a number. So the number has to be converted from being a number to
19864 being a string. This is done with the @code{number-to-string} function.
19869 (length (number-to-string 35))
19872 (length (number-to-string 100))
19878 (@code{number-to-string} is also called @code{int-to-string}; you will
19879 see this alternative name in various sources.)
19881 In addition, in each label, each number is followed by a string such
19882 as @w{@samp{ - }}, which we will call the @code{Y-axis-tic} marker.
19883 This variable is defined with @code{defvar}:
19888 (defvar Y-axis-tic " - "
19889 "String that follows number in a Y axis label.")
19893 The length of the Y label is the sum of the length of the Y axis tic
19894 mark and the length of the number of the top of the graph.
19897 (length (concat (number-to-string height) Y-axis-tic)))
19900 This value will be calculated by the @code{print-graph} function in
19901 its varlist as @code{full-Y-label-width} and passed on. (Note that we
19902 did not think to include this in the varlist when we first proposed it.)
19904 To make a complete vertical axis label, a tic mark is concatenated
19905 with a number; and the two together may be preceded by one or more
19906 spaces depending on how long the number is. The label consists of
19907 three parts: the (optional) leading spaces, the number, and the tic
19908 mark. The function is passed the value of the number for the specific
19909 row, and the value of the width of the top line, which is calculated
19910 (just once) by @code{print-graph}.
19914 (defun Y-axis-element (number full-Y-label-width)
19915 "Construct a NUMBERed label element.
19916 A numbered element looks like this ` 5 - ',
19917 and is padded as needed so all line up with
19918 the element for the largest number."
19921 (let* ((leading-spaces
19922 (- full-Y-label-width
19924 (concat (number-to-string number)
19929 (make-string leading-spaces ? )
19930 (number-to-string number)
19935 The @code{Y-axis-element} function concatenates together the leading
19936 spaces, if any; the number, as a string; and the tic mark.
19938 To figure out how many leading spaces the label will need, the
19939 function subtracts the actual length of the label---the length of the
19940 number plus the length of the tic mark---from the desired label width.
19942 @findex make-string
19943 Blank spaces are inserted using the @code{make-string} function. This
19944 function takes two arguments: the first tells it how long the string
19945 will be and the second is a symbol for the character to insert, in a
19946 special format. The format is a question mark followed by a blank
19947 space, like this, @samp{? }. @xref{Character Type, , Character Type,
19948 elisp, The GNU Emacs Lisp Reference Manual}, for a description of the
19949 syntax for characters. (Of course, you might want to replace the
19950 blank space by some other character @dots{} You know what to do.)
19952 The @code{number-to-string} function is used in the concatenation
19953 expression, to convert the number to a string that is concatenated
19954 with the leading spaces and the tic mark.
19956 @node Y-axis-column
19957 @appendixsubsec Create a Y Axis Column
19959 The preceding functions provide all the tools needed to construct a
19960 function that generates a list of numbered and blank strings to insert
19961 as the label for the vertical axis:
19963 @findex Y-axis-column
19966 (defun Y-axis-column (height width-of-label)
19967 "Construct list of Y axis labels and blank strings.
19968 For HEIGHT of line above base and WIDTH-OF-LABEL."
19972 (while (> height 1)
19973 (if (zerop (% height Y-axis-label-spacing))
19974 ;; @r{Insert label.}
19977 (Y-axis-element height width-of-label)
19981 ;; @r{Else, insert blanks.}
19984 (make-string width-of-label ? )
19986 (setq height (1- height)))
19987 ;; @r{Insert base line.}
19989 (cons (Y-axis-element 1 width-of-label) Y-axis))
19990 (nreverse Y-axis)))
19994 In this function, we start with the value of @code{height} and
19995 repetitively subtract one from its value. After each subtraction, we
19996 test to see whether the value is an integral multiple of the
19997 @code{Y-axis-label-spacing}. If it is, we construct a numbered label
19998 using the @code{Y-axis-element} function; if not, we construct a
19999 blank label using the @code{make-string} function. The base line
20000 consists of the number one followed by a tic mark.
20003 @node print-Y-axis Penultimate
20004 @appendixsubsec The Not Quite Final Version of @code{print-Y-axis}
20006 The list constructed by the @code{Y-axis-column} function is passed to
20007 the @code{print-Y-axis} function, which inserts the list as a column.
20009 @findex print-Y-axis
20012 (defun print-Y-axis (height full-Y-label-width)
20013 "Insert Y axis using HEIGHT and FULL-Y-LABEL-WIDTH.
20014 Height must be the maximum height of the graph.
20015 Full width is the width of the highest label element."
20016 ;; Value of height and full-Y-label-width
20017 ;; are passed by `print-graph'.
20020 (let ((start (point)))
20022 (Y-axis-column height full-Y-label-width))
20023 ;; @r{Place point ready for inserting graph.}
20025 ;; @r{Move point forward by value of} full-Y-label-width
20026 (forward-char full-Y-label-width)))
20030 The @code{print-Y-axis} uses the @code{insert-rectangle} function to
20031 insert the Y axis labels created by the @code{Y-axis-column} function.
20032 In addition, it places point at the correct position for printing the body of
20035 You can test @code{print-Y-axis}:
20043 Y-axis-label-spacing
20052 Copy the following expression:
20055 (print-Y-axis 12 5)
20059 Switch to the @file{*scratch*} buffer and place the cursor where you
20060 want the axis labels to start.
20063 Type @kbd{M-:} (@code{eval-expression}).
20066 Yank the @code{graph-body-print} expression into the minibuffer
20067 with @kbd{C-y} (@code{yank)}.
20070 Press @key{RET} to evaluate the expression.
20073 Emacs will print labels vertically, the top one being @w{@samp{10 -@w{
20074 }}}. (The @code{print-graph} function will pass the value of
20075 @code{height-of-top-line}, which in this case will end up as 15,
20076 thereby getting rid of what might appear as a bug.)
20080 @appendixsec The @code{print-X-axis} Function
20081 @cindex Axis, print horizontal
20082 @cindex X axis printing
20083 @cindex Print horizontal axis
20084 @cindex Horizontal axis printing
20086 X axis labels are much like Y axis labels, except that the ticks are on a
20087 line above the numbers. Labels should look like this:
20096 The first tic is under the first column of the graph and is preceded by
20097 several blank spaces. These spaces provide room in rows above for the Y
20098 axis labels. The second, third, fourth, and subsequent ticks are all
20099 spaced equally, according to the value of @code{X-axis-label-spacing}.
20101 The second row of the X axis consists of numbers, preceded by several
20102 blank spaces and also separated according to the value of the variable
20103 @code{X-axis-label-spacing}.
20105 The value of the variable @code{X-axis-label-spacing} should itself be
20106 measured in units of @code{symbol-width}, since you may want to change
20107 the width of the symbols that you are using to print the body of the
20108 graph without changing the ways the graph is labeled.
20111 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
20112 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
20116 @node Similarities differences
20117 @unnumberedsubsec Similarities and differences
20120 The @code{print-X-axis} function is constructed in more or less the
20121 same fashion as the @code{print-Y-axis} function except that it has
20122 two lines: the line of tic marks and the numbers. We will write a
20123 separate function to print each line and then combine them within the
20124 @code{print-X-axis} function.
20126 This is a three step process:
20130 Write a function to print the X axis tic marks, @code{print-X-axis-tic-line}.
20133 Write a function to print the X numbers, @code{print-X-axis-numbered-line}.
20136 Write a function to print both lines, the @code{print-X-axis} function,
20137 using @code{print-X-axis-tic-line} and
20138 @code{print-X-axis-numbered-line}.
20141 @node X Axis Tic Marks
20142 @appendixsubsec X Axis Tic Marks
20144 The first function should print the X axis tic marks. We must specify
20145 the tic marks themselves and their spacing:
20149 (defvar X-axis-label-spacing
20150 (if (boundp 'graph-blank)
20151 (* 5 (length graph-blank)) 5)
20152 "Number of units from one X axis label to next.")
20157 (Note that the value of @code{graph-blank} is set by another
20158 @code{defvar}. The @code{boundp} predicate checks whether it has
20159 already been set; @code{boundp} returns @code{nil} if it has not. If
20160 @code{graph-blank} were unbound and we did not use this conditional
20161 construction, in a recent GNU Emacs, we would enter the debugger and
20162 see an error message saying @samp{@w{Debugger entered--Lisp error:}
20163 @w{(void-variable graph-blank)}}.)
20166 Here is the @code{defvar} for @code{X-axis-tic-symbol}:
20170 (defvar X-axis-tic-symbol "|"
20171 "String to insert to point to a column in X axis.")
20176 The goal is to make a line that looks like this:
20182 The first tic is indented so that it is under the first column, which is
20183 indented to provide space for the Y axis labels.
20185 A tic element consists of the blank spaces that stretch from one tic to
20186 the next plus a tic symbol. The number of blanks is determined by the
20187 width of the tic symbol and the @code{X-axis-label-spacing}.
20190 The code looks like this:
20194 ;;; X-axis-tic-element
20198 ;; @r{Make a string of blanks.}
20199 (- (* symbol-width X-axis-label-spacing)
20200 (length X-axis-tic-symbol))
20202 ;; @r{Concatenate blanks with tic symbol.}
20208 Next, we determine how many blanks are needed to indent the first tic
20209 mark to the first column of the graph. This uses the value of
20210 @code{full-Y-label-width} passed it by the @code{print-graph} function.
20213 The code to make @code{X-axis-leading-spaces}
20218 ;; X-axis-leading-spaces
20220 (make-string full-Y-label-width ? )
20225 We also need to determine the length of the horizontal axis, which is
20226 the length of the numbers list, and the number of ticks in the horizontal
20233 (length numbers-list)
20239 (* symbol-width X-axis-label-spacing)
20243 ;; number-of-X-ticks
20244 (if (zerop (% (X-length tic-width)))
20245 (/ (X-length tic-width))
20246 (1+ (/ (X-length tic-width))))
20251 All this leads us directly to the function for printing the X axis tic line:
20253 @findex print-X-axis-tic-line
20256 (defun print-X-axis-tic-line
20257 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
20258 "Print ticks for X axis."
20259 (insert X-axis-leading-spaces)
20260 (insert X-axis-tic-symbol) ; @r{Under first column.}
20263 ;; @r{Insert second tic in the right spot.}
20266 (- (* symbol-width X-axis-label-spacing)
20267 ;; @r{Insert white space up to second tic symbol.}
20268 (* 2 (length X-axis-tic-symbol)))
20270 X-axis-tic-symbol))
20273 ;; @r{Insert remaining ticks.}
20274 (while (> number-of-X-tics 1)
20275 (insert X-axis-tic-element)
20276 (setq number-of-X-tics (1- number-of-X-tics))))
20280 The line of numbers is equally straightforward:
20283 First, we create a numbered element with blank spaces before each number:
20285 @findex X-axis-element
20288 (defun X-axis-element (number)
20289 "Construct a numbered X axis element."
20290 (let ((leading-spaces
20291 (- (* symbol-width X-axis-label-spacing)
20292 (length (number-to-string number)))))
20293 (concat (make-string leading-spaces ? )
20294 (number-to-string number))))
20298 Next, we create the function to print the numbered line, starting with
20299 the number ``1'' under the first column:
20301 @findex print-X-axis-numbered-line
20304 (defun print-X-axis-numbered-line
20305 (number-of-X-tics X-axis-leading-spaces)
20306 "Print line of X-axis numbers"
20307 (let ((number X-axis-label-spacing))
20308 (insert X-axis-leading-spaces)
20314 ;; @r{Insert white space up to next number.}
20315 (- (* symbol-width X-axis-label-spacing) 2)
20317 (number-to-string number)))
20320 ;; @r{Insert remaining numbers.}
20321 (setq number (+ number X-axis-label-spacing))
20322 (while (> number-of-X-tics 1)
20323 (insert (X-axis-element number))
20324 (setq number (+ number X-axis-label-spacing))
20325 (setq number-of-X-tics (1- number-of-X-tics)))))
20329 Finally, we need to write the @code{print-X-axis} that uses
20330 @code{print-X-axis-tic-line} and
20331 @code{print-X-axis-numbered-line}.
20333 The function must determine the local values of the variables used by both
20334 @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}, and
20335 then it must call them. Also, it must print the carriage return that
20336 separates the two lines.
20338 The function consists of a varlist that specifies five local variables,
20339 and calls to each of the two line printing functions:
20341 @findex print-X-axis
20344 (defun print-X-axis (numbers-list)
20345 "Print X axis labels to length of NUMBERS-LIST."
20346 (let* ((leading-spaces
20347 (make-string full-Y-label-width ? ))
20350 ;; symbol-width @r{is provided by} graph-body-print
20351 (tic-width (* symbol-width X-axis-label-spacing))
20352 (X-length (length numbers-list))
20360 ;; @r{Make a string of blanks.}
20361 (- (* symbol-width X-axis-label-spacing)
20362 (length X-axis-tic-symbol))
20366 ;; @r{Concatenate blanks with tic symbol.}
20367 X-axis-tic-symbol))
20371 (if (zerop (% X-length tic-width))
20372 (/ X-length tic-width)
20373 (1+ (/ X-length tic-width)))))
20376 (print-X-axis-tic-line tic-number leading-spaces X-tic)
20378 (print-X-axis-numbered-line tic-number leading-spaces)))
20383 You can test @code{print-X-axis}:
20387 Install @code{X-axis-tic-symbol}, @code{X-axis-label-spacing},
20388 @code{print-X-axis-tic-line}, as well as @code{X-axis-element},
20389 @code{print-X-axis-numbered-line}, and @code{print-X-axis}.
20392 Copy the following expression:
20397 (let ((full-Y-label-width 5)
20400 '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16))))
20405 Switch to the @file{*scratch*} buffer and place the cursor where you
20406 want the axis labels to start.
20409 Type @kbd{M-:} (@code{eval-expression}).
20412 Yank the test expression into the minibuffer
20413 with @kbd{C-y} (@code{yank)}.
20416 Press @key{RET} to evaluate the expression.
20420 Emacs will print the horizontal axis like this:
20430 @node Print Whole Graph
20431 @appendixsec Printing the Whole Graph
20432 @cindex Printing the whole graph
20433 @cindex Whole graph printing
20434 @cindex Graph, printing all
20436 Now we are nearly ready to print the whole graph.
20438 The function to print the graph with the proper labels follows the
20439 outline we created earlier (@pxref{Full Graph, , A Graph with Labeled
20440 Axes}), but with additions.
20443 Here is the outline:
20447 (defun print-graph (numbers-list)
20448 "@var{documentation}@dots{}"
20449 (let ((height @dots{}
20453 (print-Y-axis height @dots{} )
20454 (graph-body-print numbers-list)
20455 (print-X-axis @dots{} )))
20460 * The final version:: A few changes.
20461 * Test print-graph:: Run a short test.
20462 * Graphing words in defuns:: Executing the final code.
20463 * lambda:: How to write an anonymous function.
20464 * mapcar:: Apply a function to elements of a list.
20465 * Another Bug:: Yet another bug @dots{} most insidious.
20466 * Final printed graph:: The graph itself!
20470 @node The final version
20471 @unnumberedsubsec Changes for the Final Version
20474 The final version is different from what we planned in two ways:
20475 first, it contains additional values calculated once in the varlist;
20476 second, it carries an option to specify the labels' increment per row.
20477 This latter feature turns out to be essential; otherwise, a graph may
20478 have more rows than fit on a display or on a sheet of paper.
20481 This new feature requires a change to the @code{Y-axis-column}
20482 function, to add @code{vertical-step} to it. The function looks like
20485 @findex Y-axis-column @r{Final version.}
20488 ;;; @r{Final version.}
20489 (defun Y-axis-column
20490 (height width-of-label &optional vertical-step)
20491 "Construct list of labels for Y axis.
20492 HEIGHT is maximum height of graph.
20493 WIDTH-OF-LABEL is maximum width of label.
20494 VERTICAL-STEP, an option, is a positive integer
20495 that specifies how much a Y axis label increments
20496 for each line. For example, a step of 5 means
20497 that each line is five units of the graph."
20501 (number-per-line (or vertical-step 1)))
20502 (while (> height 1)
20503 (if (zerop (% height Y-axis-label-spacing))
20506 ;; @r{Insert label.}
20510 (* height number-per-line)
20515 ;; @r{Else, insert blanks.}
20518 (make-string width-of-label ? )
20520 (setq height (1- height)))
20523 ;; @r{Insert base line.}
20524 (setq Y-axis (cons (Y-axis-element
20525 (or vertical-step 1)
20528 (nreverse Y-axis)))
20532 The values for the maximum height of graph and the width of a symbol
20533 are computed by @code{print-graph} in its @code{let} expression; so
20534 @code{graph-body-print} must be changed to accept them.
20536 @findex graph-body-print @r{Final version.}
20539 ;;; @r{Final version.}
20540 (defun graph-body-print (numbers-list height symbol-width)
20541 "Print a bar graph of the NUMBERS-LIST.
20542 The numbers-list consists of the Y-axis values.
20543 HEIGHT is maximum height of graph.
20544 SYMBOL-WIDTH is number of each column."
20547 (let (from-position)
20548 (while numbers-list
20549 (setq from-position (point))
20551 (column-of-graph height (car numbers-list)))
20552 (goto-char from-position)
20553 (forward-char symbol-width)
20556 ;; @r{Draw graph column by column.}
20558 (setq numbers-list (cdr numbers-list)))
20559 ;; @r{Place point for X axis labels.}
20560 (forward-line height)
20566 Finally, the code for the @code{print-graph} function:
20568 @findex print-graph @r{Final version.}
20571 ;;; @r{Final version.}
20573 (numbers-list &optional vertical-step)
20574 "Print labeled bar graph of the NUMBERS-LIST.
20575 The numbers-list consists of the Y-axis values.
20579 Optionally, VERTICAL-STEP, a positive integer,
20580 specifies how much a Y axis label increments for
20581 each line. For example, a step of 5 means that
20582 each row is five units."
20585 (let* ((symbol-width (length graph-blank))
20586 ;; @code{height} @r{is both the largest number}
20587 ;; @r{and the number with the most digits.}
20588 (height (apply 'max numbers-list))
20591 (height-of-top-line
20592 (if (zerop (% height Y-axis-label-spacing))
20595 (* (1+ (/ height Y-axis-label-spacing))
20596 Y-axis-label-spacing)))
20599 (vertical-step (or vertical-step 1))
20600 (full-Y-label-width
20606 (* height-of-top-line vertical-step))
20612 height-of-top-line full-Y-label-width vertical-step)
20616 numbers-list height-of-top-line symbol-width)
20617 (print-X-axis numbers-list)))
20621 @node Test print-graph
20622 @appendixsubsec Testing @code{print-graph}
20625 We can test the @code{print-graph} function with a short list of numbers:
20629 Install the final versions of @code{Y-axis-column},
20630 @code{graph-body-print}, and @code{print-graph} (in addition to the
20634 Copy the following expression:
20637 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1))
20641 Switch to the @file{*scratch*} buffer and place the cursor where you
20642 want the axis labels to start.
20645 Type @kbd{M-:} (@code{eval-expression}).
20648 Yank the test expression into the minibuffer
20649 with @kbd{C-y} (@code{yank)}.
20652 Press @key{RET} to evaluate the expression.
20656 Emacs will print a graph that looks like this:
20677 On the other hand, if you pass @code{print-graph} a
20678 @code{vertical-step} value of 2, by evaluating this expression:
20681 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1) 2)
20686 The graph looks like this:
20707 (A question: is the `2' on the bottom of the vertical axis a bug or a
20708 feature? If you think it is a bug, and should be a `1' instead, (or
20709 even a `0'), you can modify the sources.)
20711 @node Graphing words in defuns
20712 @appendixsubsec Graphing Numbers of Words and Symbols
20714 Now for the graph for which all this code was written: a graph that
20715 shows how many function definitions contain fewer than 10 words and
20716 symbols, how many contain between 10 and 19 words and symbols, how
20717 many contain between 20 and 29 words and symbols, and so on.
20719 This is a multi-step process. First make sure you have loaded all the
20723 It is a good idea to reset the value of @code{top-of-ranges} in case
20724 you have set it to some different value. You can evaluate the
20729 (setq top-of-ranges
20732 110 120 130 140 150
20733 160 170 180 190 200
20734 210 220 230 240 250
20735 260 270 280 290 300)
20740 Next create a list of the number of words and symbols in each range.
20744 Evaluate the following:
20748 (setq list-for-graph
20751 (recursive-lengths-list-many-files
20752 (directory-files "/usr/local/emacs/lisp"
20760 On my old machine, this took about an hour. It looked though 303 Lisp
20761 files in my copy of Emacs version 19.23. After all that computing,
20762 the @code{list-for-graph} had this value:
20766 (537 1027 955 785 594 483 349 292 224 199 166 120 116 99
20767 90 80 67 48 52 45 41 33 28 26 25 20 12 28 11 13 220)
20772 This means that my copy of Emacs had 537 function definitions with
20773 fewer than 10 words or symbols in them, 1,027 function definitions
20774 with 10 to 19 words or symbols in them, 955 function definitions with
20775 20 to 29 words or symbols in them, and so on.
20777 Clearly, just by looking at this list we can see that most function
20778 definitions contain ten to thirty words and symbols.
20780 Now for printing. We do @emph{not} want to print a graph that is
20781 1,030 lines high @dots{} Instead, we should print a graph that is
20782 fewer than twenty-five lines high. A graph that height can be
20783 displayed on almost any monitor, and easily printed on a sheet of paper.
20785 This means that each value in @code{list-for-graph} must be reduced to
20786 one-fiftieth its present value.
20788 Here is a short function to do just that, using two functions we have
20789 not yet seen, @code{mapcar} and @code{lambda}.
20793 (defun one-fiftieth (full-range)
20794 "Return list, each number one-fiftieth of previous."
20795 (mapcar (lambda (arg) (/ arg 50)) full-range))
20800 @appendixsubsec A @code{lambda} Expression: Useful Anonymity
20801 @cindex Anonymous function
20804 @code{lambda} is the symbol for an anonymous function, a function
20805 without a name. Every time you use an anonymous function, you need to
20806 include its whole body.
20813 (lambda (arg) (/ arg 50))
20817 is a function definition that says `return the value resulting from
20818 dividing whatever is passed to me as @code{arg} by 50'.
20821 Earlier, for example, we had a function @code{multiply-by-seven}; it
20822 multiplied its argument by 7. This function is similar, except it
20823 divides its argument by 50; and, it has no name. The anonymous
20824 equivalent of @code{multiply-by-seven} is:
20827 (lambda (number) (* 7 number))
20831 (@xref{defun, , The @code{defun} Macro}.)
20835 If we want to multiply 3 by 7, we can write:
20837 @c clear print-postscript-figures
20838 @c lambda example diagram #1
20842 (multiply-by-seven 3)
20843 \_______________/ ^
20849 @ifset print-postscript-figures
20852 @center @image{lambda-1}
20856 @ifclear print-postscript-figures
20860 (multiply-by-seven 3)
20861 \_______________/ ^
20870 This expression returns 21.
20874 Similarly, we can write:
20876 @c lambda example diagram #2
20880 ((lambda (number) (* 7 number)) 3)
20881 \____________________________/ ^
20883 anonymous function argument
20887 @ifset print-postscript-figures
20890 @center @image{lambda-2}
20894 @ifclear print-postscript-figures
20898 ((lambda (number) (* 7 number)) 3)
20899 \____________________________/ ^
20901 anonymous function argument
20909 If we want to divide 100 by 50, we can write:
20911 @c lambda example diagram #3
20915 ((lambda (arg) (/ arg 50)) 100)
20916 \______________________/ \_/
20918 anonymous function argument
20922 @ifset print-postscript-figures
20925 @center @image{lambda-3}
20929 @ifclear print-postscript-figures
20933 ((lambda (arg) (/ arg 50)) 100)
20934 \______________________/ \_/
20936 anonymous function argument
20943 This expression returns 2. The 100 is passed to the function, which
20944 divides that number by 50.
20946 @xref{Lambda Expressions, , Lambda Expressions, elisp, The GNU Emacs
20947 Lisp Reference Manual}, for more about @code{lambda}. Lisp and lambda
20948 expressions derive from the Lambda Calculus.
20951 @appendixsubsec The @code{mapcar} Function
20954 @code{mapcar} is a function that calls its first argument with each
20955 element of its second argument, in turn. The second argument must be
20958 The @samp{map} part of the name comes from the mathematical phrase,
20959 `mapping over a domain', meaning to apply a function to each of the
20960 elements in a domain. The mathematical phrase is based on the
20961 metaphor of a surveyor walking, one step at a time, over an area he is
20962 mapping. And @samp{car}, of course, comes from the Lisp notion of the
20971 (mapcar '1+ '(2 4 6))
20977 The function @code{1+} which adds one to its argument, is executed on
20978 @emph{each} element of the list, and a new list is returned.
20980 Contrast this with @code{apply}, which applies its first argument to
20982 (@xref{Readying a Graph, , Readying a Graph}, for a explanation of
20986 In the definition of @code{one-fiftieth}, the first argument is the
20987 anonymous function:
20990 (lambda (arg) (/ arg 50))
20994 and the second argument is @code{full-range}, which will be bound to
20995 @code{list-for-graph}.
20998 The whole expression looks like this:
21001 (mapcar (lambda (arg) (/ arg 50)) full-range))
21004 @xref{Mapping Functions, , Mapping Functions, elisp, The GNU Emacs
21005 Lisp Reference Manual}, for more about @code{mapcar}.
21007 Using the @code{one-fiftieth} function, we can generate a list in
21008 which each element is one-fiftieth the size of the corresponding
21009 element in @code{list-for-graph}.
21013 (setq fiftieth-list-for-graph
21014 (one-fiftieth list-for-graph))
21019 The resulting list looks like this:
21023 (10 20 19 15 11 9 6 5 4 3 3 2 2
21024 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 4)
21029 This, we are almost ready to print! (We also notice the loss of
21030 information: many of the higher ranges are 0, meaning that fewer than
21031 50 defuns had that many words or symbols---but not necessarily meaning
21032 that none had that many words or symbols.)
21035 @appendixsubsec Another Bug @dots{} Most Insidious
21036 @cindex Bug, most insidious type
21037 @cindex Insidious type of bug
21039 I said `almost ready to print'! Of course, there is a bug in the
21040 @code{print-graph} function @dots{} It has a @code{vertical-step}
21041 option, but not a @code{horizontal-step} option. The
21042 @code{top-of-range} scale goes from 10 to 300 by tens. But the
21043 @code{print-graph} function will print only by ones.
21045 This is a classic example of what some consider the most insidious
21046 type of bug, the bug of omission. This is not the kind of bug you can
21047 find by studying the code, for it is not in the code; it is an omitted
21048 feature. Your best actions are to try your program early and often;
21049 and try to arrange, as much as you can, to write code that is easy to
21050 understand and easy to change. Try to be aware, whenever you can,
21051 that whatever you have written, @emph{will} be rewritten, if not soon,
21052 eventually. A hard maxim to follow.
21054 It is the @code{print-X-axis-numbered-line} function that needs the
21055 work; and then the @code{print-X-axis} and the @code{print-graph}
21056 functions need to be adapted. Not much needs to be done; there is one
21057 nicety: the numbers ought to line up under the tic marks. This takes
21061 Here is the corrected @code{print-X-axis-numbered-line}:
21065 (defun print-X-axis-numbered-line
21066 (number-of-X-tics X-axis-leading-spaces
21067 &optional horizontal-step)
21068 "Print line of X-axis numbers"
21069 (let ((number X-axis-label-spacing)
21070 (horizontal-step (or horizontal-step 1)))
21073 (insert X-axis-leading-spaces)
21074 ;; @r{Delete extra leading spaces.}
21077 (length (number-to-string horizontal-step)))))
21082 ;; @r{Insert white space.}
21084 X-axis-label-spacing)
21087 (number-to-string horizontal-step)))
21091 (* number horizontal-step))))
21094 ;; @r{Insert remaining numbers.}
21095 (setq number (+ number X-axis-label-spacing))
21096 (while (> number-of-X-tics 1)
21097 (insert (X-axis-element
21098 (* number horizontal-step)))
21099 (setq number (+ number X-axis-label-spacing))
21100 (setq number-of-X-tics (1- number-of-X-tics)))))
21105 If you are reading this in Info, you can see the new versions of
21106 @code{print-X-axis} @code{print-graph} and evaluate them. If you are
21107 reading this in a printed book, you can see the changed lines here
21108 (the full text is too much to print).
21113 (defun print-X-axis (numbers-list horizontal-step)
21115 (print-X-axis-numbered-line
21116 tic-number leading-spaces horizontal-step))
21124 &optional vertical-step horizontal-step)
21126 (print-X-axis numbers-list horizontal-step))
21134 (defun print-X-axis (numbers-list horizontal-step)
21135 "Print X axis labels to length of NUMBERS-LIST.
21136 Optionally, HORIZONTAL-STEP, a positive integer,
21137 specifies how much an X axis label increments for
21141 ;; Value of symbol-width and full-Y-label-width
21142 ;; are passed by `print-graph'.
21143 (let* ((leading-spaces
21144 (make-string full-Y-label-width ? ))
21145 ;; symbol-width @r{is provided by} graph-body-print
21146 (tic-width (* symbol-width X-axis-label-spacing))
21147 (X-length (length numbers-list))
21153 ;; @r{Make a string of blanks.}
21154 (- (* symbol-width X-axis-label-spacing)
21155 (length X-axis-tic-symbol))
21159 ;; @r{Concatenate blanks with tic symbol.}
21160 X-axis-tic-symbol))
21162 (if (zerop (% X-length tic-width))
21163 (/ X-length tic-width)
21164 (1+ (/ X-length tic-width)))))
21168 (print-X-axis-tic-line
21169 tic-number leading-spaces X-tic)
21171 (print-X-axis-numbered-line
21172 tic-number leading-spaces horizontal-step)))
21179 (numbers-list &optional vertical-step horizontal-step)
21180 "Print labeled bar graph of the NUMBERS-LIST.
21181 The numbers-list consists of the Y-axis values.
21185 Optionally, VERTICAL-STEP, a positive integer,
21186 specifies how much a Y axis label increments for
21187 each line. For example, a step of 5 means that
21188 each row is five units.
21192 Optionally, HORIZONTAL-STEP, a positive integer,
21193 specifies how much an X axis label increments for
21195 (let* ((symbol-width (length graph-blank))
21196 ;; @code{height} @r{is both the largest number}
21197 ;; @r{and the number with the most digits.}
21198 (height (apply 'max numbers-list))
21201 (height-of-top-line
21202 (if (zerop (% height Y-axis-label-spacing))
21205 (* (1+ (/ height Y-axis-label-spacing))
21206 Y-axis-label-spacing)))
21209 (vertical-step (or vertical-step 1))
21210 (full-Y-label-width
21214 (* height-of-top-line vertical-step))
21219 height-of-top-line full-Y-label-width vertical-step)
21221 numbers-list height-of-top-line symbol-width)
21222 (print-X-axis numbers-list horizontal-step)))
21229 Graphing Definitions Re-listed
21232 Here are all the graphing definitions in their final form:
21236 (defvar top-of-ranges
21239 110 120 130 140 150
21240 160 170 180 190 200
21241 210 220 230 240 250)
21242 "List specifying ranges for `defuns-per-range'.")
21246 (defvar graph-symbol "*"
21247 "String used as symbol in graph, usually an asterisk.")
21251 (defvar graph-blank " "
21252 "String used as blank in graph, usually a blank space.
21253 graph-blank must be the same number of columns wide
21258 (defvar Y-axis-tic " - "
21259 "String that follows number in a Y axis label.")
21263 (defvar Y-axis-label-spacing 5
21264 "Number of lines from one Y axis label to next.")
21268 (defvar X-axis-tic-symbol "|"
21269 "String to insert to point to a column in X axis.")
21273 (defvar X-axis-label-spacing
21274 (if (boundp 'graph-blank)
21275 (* 5 (length graph-blank)) 5)
21276 "Number of units from one X axis label to next.")
21282 (defun count-words-in-defun ()
21283 "Return the number of words and symbols in a defun."
21284 (beginning-of-defun)
21286 (end (save-excursion (end-of-defun) (point))))
21291 (and (< (point) end)
21293 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
21295 (setq count (1+ count)))
21302 (defun lengths-list-file (filename)
21303 "Return list of definitions' lengths within FILE.
21304 The returned list is a list of numbers.
21305 Each number is the number of words or
21306 symbols in one function definition."
21310 (message "Working on `%s' ... " filename)
21312 (let ((buffer (find-file-noselect filename))
21314 (set-buffer buffer)
21315 (setq buffer-read-only t)
21317 (goto-char (point-min))
21321 (while (re-search-forward "^(defun" nil t)
21323 (cons (count-words-in-defun) lengths-list)))
21324 (kill-buffer buffer)
21331 (defun lengths-list-many-files (list-of-files)
21332 "Return list of lengths of defuns in LIST-OF-FILES."
21333 (let (lengths-list)
21334 ;;; @r{true-or-false-test}
21335 (while list-of-files
21341 ;;; @r{Generate a lengths' list.}
21343 (expand-file-name (car list-of-files)))))
21344 ;;; @r{Make files' list shorter.}
21345 (setq list-of-files (cdr list-of-files)))
21346 ;;; @r{Return final value of lengths' list.}
21353 (defun defuns-per-range (sorted-lengths top-of-ranges)
21354 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
21355 (let ((top-of-range (car top-of-ranges))
21356 (number-within-range 0)
21357 defuns-per-range-list)
21362 (while top-of-ranges
21366 ;; @r{Need number for numeric test.}
21367 (car sorted-lengths)
21368 (< (car sorted-lengths) top-of-range))
21370 ;; @r{Count number of definitions within current range.}
21371 (setq number-within-range (1+ number-within-range))
21372 (setq sorted-lengths (cdr sorted-lengths)))
21376 ;; @r{Exit inner loop but remain within outer loop.}
21378 (setq defuns-per-range-list
21379 (cons number-within-range defuns-per-range-list))
21380 (setq number-within-range 0) ; @r{Reset count to zero.}
21382 ;; @r{Move to next range.}
21383 (setq top-of-ranges (cdr top-of-ranges))
21384 ;; @r{Specify next top of range value.}
21385 (setq top-of-range (car top-of-ranges)))
21389 ;; @r{Exit outer loop and count the number of defuns larger than}
21390 ;; @r{ the largest top-of-range value.}
21391 (setq defuns-per-range-list
21393 (length sorted-lengths)
21394 defuns-per-range-list))
21396 ;; @r{Return a list of the number of definitions within each range,}
21397 ;; @r{ smallest to largest.}
21398 (nreverse defuns-per-range-list)))
21404 (defun column-of-graph (max-graph-height actual-height)
21405 "Return list of MAX-GRAPH-HEIGHT strings;
21406 ACTUAL-HEIGHT are graph-symbols.
21407 The graph-symbols are contiguous entries at the end
21409 The list will be inserted as one column of a graph.
21410 The strings are either graph-blank or graph-symbol."
21414 (let ((insert-list nil)
21415 (number-of-top-blanks
21416 (- max-graph-height actual-height)))
21418 ;; @r{Fill in @code{graph-symbols}.}
21419 (while (> actual-height 0)
21420 (setq insert-list (cons graph-symbol insert-list))
21421 (setq actual-height (1- actual-height)))
21425 ;; @r{Fill in @code{graph-blanks}.}
21426 (while (> number-of-top-blanks 0)
21427 (setq insert-list (cons graph-blank insert-list))
21428 (setq number-of-top-blanks
21429 (1- number-of-top-blanks)))
21431 ;; @r{Return whole list.}
21438 (defun Y-axis-element (number full-Y-label-width)
21439 "Construct a NUMBERed label element.
21440 A numbered element looks like this ` 5 - ',
21441 and is padded as needed so all line up with
21442 the element for the largest number."
21445 (let* ((leading-spaces
21446 (- full-Y-label-width
21448 (concat (number-to-string number)
21453 (make-string leading-spaces ? )
21454 (number-to-string number)
21461 (defun print-Y-axis
21462 (height full-Y-label-width &optional vertical-step)
21463 "Insert Y axis by HEIGHT and FULL-Y-LABEL-WIDTH.
21464 Height must be the maximum height of the graph.
21465 Full width is the width of the highest label element.
21466 Optionally, print according to VERTICAL-STEP."
21469 ;; Value of height and full-Y-label-width
21470 ;; are passed by `print-graph'.
21471 (let ((start (point)))
21473 (Y-axis-column height full-Y-label-width vertical-step))
21476 ;; @r{Place point ready for inserting graph.}
21478 ;; @r{Move point forward by value of} full-Y-label-width
21479 (forward-char full-Y-label-width)))
21485 (defun print-X-axis-tic-line
21486 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
21487 "Print ticks for X axis."
21488 (insert X-axis-leading-spaces)
21489 (insert X-axis-tic-symbol) ; @r{Under first column.}
21492 ;; @r{Insert second tic in the right spot.}
21495 (- (* symbol-width X-axis-label-spacing)
21496 ;; @r{Insert white space up to second tic symbol.}
21497 (* 2 (length X-axis-tic-symbol)))
21499 X-axis-tic-symbol))
21502 ;; @r{Insert remaining ticks.}
21503 (while (> number-of-X-tics 1)
21504 (insert X-axis-tic-element)
21505 (setq number-of-X-tics (1- number-of-X-tics))))
21511 (defun X-axis-element (number)
21512 "Construct a numbered X axis element."
21513 (let ((leading-spaces
21514 (- (* symbol-width X-axis-label-spacing)
21515 (length (number-to-string number)))))
21516 (concat (make-string leading-spaces ? )
21517 (number-to-string number))))
21523 (defun graph-body-print (numbers-list height symbol-width)
21524 "Print a bar graph of the NUMBERS-LIST.
21525 The numbers-list consists of the Y-axis values.
21526 HEIGHT is maximum height of graph.
21527 SYMBOL-WIDTH is number of each column."
21530 (let (from-position)
21531 (while numbers-list
21532 (setq from-position (point))
21534 (column-of-graph height (car numbers-list)))
21535 (goto-char from-position)
21536 (forward-char symbol-width)
21539 ;; @r{Draw graph column by column.}
21541 (setq numbers-list (cdr numbers-list)))
21542 ;; @r{Place point for X axis labels.}
21543 (forward-line height)
21550 (defun Y-axis-column
21551 (height width-of-label &optional vertical-step)
21552 "Construct list of labels for Y axis.
21553 HEIGHT is maximum height of graph.
21554 WIDTH-OF-LABEL is maximum width of label.
21557 VERTICAL-STEP, an option, is a positive integer
21558 that specifies how much a Y axis label increments
21559 for each line. For example, a step of 5 means
21560 that each line is five units of the graph."
21562 (number-per-line (or vertical-step 1)))
21565 (while (> height 1)
21566 (if (zerop (% height Y-axis-label-spacing))
21567 ;; @r{Insert label.}
21571 (* height number-per-line)
21576 ;; @r{Else, insert blanks.}
21579 (make-string width-of-label ? )
21581 (setq height (1- height)))
21584 ;; @r{Insert base line.}
21585 (setq Y-axis (cons (Y-axis-element
21586 (or vertical-step 1)
21589 (nreverse Y-axis)))
21595 (defun print-X-axis-numbered-line
21596 (number-of-X-tics X-axis-leading-spaces
21597 &optional horizontal-step)
21598 "Print line of X-axis numbers"
21599 (let ((number X-axis-label-spacing)
21600 (horizontal-step (or horizontal-step 1)))
21603 (insert X-axis-leading-spaces)
21605 (delete-char (- (1- (length (number-to-string horizontal-step)))))
21608 ;; @r{Insert white space up to next number.}
21609 (- (* symbol-width X-axis-label-spacing)
21610 (1- (length (number-to-string horizontal-step)))
21613 (number-to-string (* number horizontal-step))))
21616 ;; @r{Insert remaining numbers.}
21617 (setq number (+ number X-axis-label-spacing))
21618 (while (> number-of-X-tics 1)
21619 (insert (X-axis-element (* number horizontal-step)))
21620 (setq number (+ number X-axis-label-spacing))
21621 (setq number-of-X-tics (1- number-of-X-tics)))))
21627 (defun print-X-axis (numbers-list horizontal-step)
21628 "Print X axis labels to length of NUMBERS-LIST.
21629 Optionally, HORIZONTAL-STEP, a positive integer,
21630 specifies how much an X axis label increments for
21634 ;; Value of symbol-width and full-Y-label-width
21635 ;; are passed by `print-graph'.
21636 (let* ((leading-spaces
21637 (make-string full-Y-label-width ? ))
21638 ;; symbol-width @r{is provided by} graph-body-print
21639 (tic-width (* symbol-width X-axis-label-spacing))
21640 (X-length (length numbers-list))
21646 ;; @r{Make a string of blanks.}
21647 (- (* symbol-width X-axis-label-spacing)
21648 (length X-axis-tic-symbol))
21652 ;; @r{Concatenate blanks with tic symbol.}
21653 X-axis-tic-symbol))
21655 (if (zerop (% X-length tic-width))
21656 (/ X-length tic-width)
21657 (1+ (/ X-length tic-width)))))
21661 (print-X-axis-tic-line
21662 tic-number leading-spaces X-tic)
21664 (print-X-axis-numbered-line
21665 tic-number leading-spaces horizontal-step)))
21671 (defun one-fiftieth (full-range)
21672 "Return list, each number of which is 1/50th previous."
21673 (mapcar (lambda (arg) (/ arg 50)) full-range))
21680 (numbers-list &optional vertical-step horizontal-step)
21681 "Print labeled bar graph of the NUMBERS-LIST.
21682 The numbers-list consists of the Y-axis values.
21686 Optionally, VERTICAL-STEP, a positive integer,
21687 specifies how much a Y axis label increments for
21688 each line. For example, a step of 5 means that
21689 each row is five units.
21693 Optionally, HORIZONTAL-STEP, a positive integer,
21694 specifies how much an X axis label increments for
21696 (let* ((symbol-width (length graph-blank))
21697 ;; @code{height} @r{is both the largest number}
21698 ;; @r{and the number with the most digits.}
21699 (height (apply 'max numbers-list))
21702 (height-of-top-line
21703 (if (zerop (% height Y-axis-label-spacing))
21706 (* (1+ (/ height Y-axis-label-spacing))
21707 Y-axis-label-spacing)))
21710 (vertical-step (or vertical-step 1))
21711 (full-Y-label-width
21715 (* height-of-top-line vertical-step))
21721 height-of-top-line full-Y-label-width vertical-step)
21723 numbers-list height-of-top-line symbol-width)
21724 (print-X-axis numbers-list horizontal-step)))
21731 @node Final printed graph
21732 @appendixsubsec The Printed Graph
21734 When made and installed, you can call the @code{print-graph} command
21740 (print-graph fiftieth-list-for-graph 50 10)
21770 50 - ***************** * *
21772 10 50 100 150 200 250 300 350
21779 The largest group of functions contain 10--19 words and symbols each.
21781 @node Free Software and Free Manuals
21782 @appendix Free Software and Free Manuals
21784 @strong{by Richard M. Stallman}
21787 The biggest deficiency in free operating systems is not in the
21788 software---it is the lack of good free manuals that we can include in
21789 these systems. Many of our most important programs do not come with
21790 full manuals. Documentation is an essential part of any software
21791 package; when an important free software package does not come with a
21792 free manual, that is a major gap. We have many such gaps today.
21794 Once upon a time, many years ago, I thought I would learn Perl. I got
21795 a copy of a free manual, but I found it hard to read. When I asked
21796 Perl users about alternatives, they told me that there were better
21797 introductory manuals---but those were not free.
21799 Why was this? The authors of the good manuals had written them for
21800 O'Reilly Associates, which published them with restrictive terms---no
21801 copying, no modification, source files not available---which exclude
21802 them from the free software community.
21804 That wasn't the first time this sort of thing has happened, and (to
21805 our community's great loss) it was far from the last. Proprietary
21806 manual publishers have enticed a great many authors to restrict their
21807 manuals since then. Many times I have heard a GNU user eagerly tell me
21808 about a manual that he is writing, with which he expects to help the
21809 GNU project---and then had my hopes dashed, as he proceeded to explain
21810 that he had signed a contract with a publisher that would restrict it
21811 so that we cannot use it.
21813 Given that writing good English is a rare skill among programmers, we
21814 can ill afford to lose manuals this way.
21816 Free documentation, like free software, is a matter of freedom, not
21817 price. The problem with these manuals was not that O'Reilly Associates
21818 charged a price for printed copies---that in itself is fine. The Free
21819 Software Foundation @uref{http://shop.fsf.org, sells printed copies} of
21820 free @uref{http://www.gnu.org/doc/doc.html, GNU manuals}, too.
21821 But GNU manuals are available in source code form, while these manuals
21822 are available only on paper. GNU manuals come with permission to copy
21823 and modify; the Perl manuals do not. These restrictions are the
21826 The criterion for a free manual is pretty much the same as for free
21827 software: it is a matter of giving all users certain
21828 freedoms. Redistribution (including commercial redistribution) must be
21829 permitted, so that the manual can accompany every copy of the program,
21830 on-line or on paper. Permission for modification is crucial too.
21832 As a general rule, I don't believe that it is essential for people to
21833 have permission to modify all sorts of articles and books. The issues
21834 for writings are not necessarily the same as those for software. For
21835 example, I don't think you or I are obliged to give permission to
21836 modify articles like this one, which describe our actions and our
21839 But there is a particular reason why the freedom to modify is crucial
21840 for documentation for free software. When people exercise their right
21841 to modify the software, and add or change its features, if they are
21842 conscientious they will change the manual too---so they can provide
21843 accurate and usable documentation with the modified program. A manual
21844 which forbids programmers to be conscientious and finish the job, or
21845 more precisely requires them to write a new manual from scratch if
21846 they change the program, does not fill our community's needs.
21848 While a blanket prohibition on modification is unacceptable, some
21849 kinds of limits on the method of modification pose no problem. For
21850 example, requirements to preserve the original author's copyright
21851 notice, the distribution terms, or the list of authors, are ok. It is
21852 also no problem to require modified versions to include notice that
21853 they were modified, even to have entire sections that may not be
21854 deleted or changed, as long as these sections deal with nontechnical
21855 topics. (Some GNU manuals have them.)
21857 These kinds of restrictions are not a problem because, as a practical
21858 matter, they don't stop the conscientious programmer from adapting the
21859 manual to fit the modified program. In other words, they don't block
21860 the free software community from making full use of the manual.
21862 However, it must be possible to modify all the technical content of
21863 the manual, and then distribute the result in all the usual media,
21864 through all the usual channels; otherwise, the restrictions do block
21865 the community, the manual is not free, and so we need another manual.
21867 Unfortunately, it is often hard to find someone to write another
21868 manual when a proprietary manual exists. The obstacle is that many
21869 users think that a proprietary manual is good enough---so they don't
21870 see the need to write a free manual. They do not see that the free
21871 operating system has a gap that needs filling.
21873 Why do users think that proprietary manuals are good enough? Some have
21874 not considered the issue. I hope this article will do something to
21877 Other users consider proprietary manuals acceptable for the same
21878 reason so many people consider proprietary software acceptable: they
21879 judge in purely practical terms, not using freedom as a
21880 criterion. These people are entitled to their opinions, but since
21881 those opinions spring from values which do not include freedom, they
21882 are no guide for those of us who do value freedom.
21884 Please spread the word about this issue. We continue to lose manuals
21885 to proprietary publishing. If we spread the word that proprietary
21886 manuals are not sufficient, perhaps the next person who wants to help
21887 GNU by writing documentation will realize, before it is too late, that
21888 he must above all make it free.
21890 We can also encourage commercial publishers to sell free, copylefted
21891 manuals instead of proprietary ones. One way you can help this is to
21892 check the distribution terms of a manual before you buy it, and prefer
21893 copylefted manuals to non-copylefted ones.
21897 Note: The Free Software Foundation maintains a page on its Web site
21898 that lists free books available from other publishers:@*
21899 @uref{http://www.gnu.org/doc/other-free-books.html}
21901 @node GNU Free Documentation License
21902 @appendix GNU Free Documentation License
21904 @cindex FDL, GNU Free Documentation License
21905 @include doclicense.texi
21911 MENU ENTRY: NODE NAME.
21917 @c Place biographical information on right-hand (verso) page
21920 \par\vfill\supereject
21922 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21923 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21926 % \par\vfill\supereject
21927 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21928 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21929 %\page\hbox{}%\page
21930 %\page\hbox{}%\page
21937 @c ================ Biographical information ================
21941 @center About the Author
21946 @node About the Author
21947 @unnumbered About the Author
21951 Robert J. Chassell has worked with GNU Emacs since 1985. He writes
21952 and edits, teaches Emacs and Emacs Lisp, and speaks throughout the
21953 world on software freedom. Chassell was a founding Director and
21954 Treasurer of the Free Software Foundation, Inc. He is co-author of
21955 the @cite{Texinfo} manual, and has edited more than a dozen other
21956 books. He graduated from Cambridge University, in England. He has an
21957 abiding interest in social and economic history and flies his own
21964 @c @c Prevent page number on blank verso, so eject it first.
21966 @c \par\vfill\supereject
21971 @c @evenheading @thispage @| @| @thistitle
21972 @c @oddheading @| @| @thispage