Commit | Line | Data |
---|---|---|
b8d4c8d0 GM |
1 | @c -*-texinfo-*- |
2 | @c This is part of the GNU Emacs Lisp Reference Manual. | |
3 | @c Copyright (C) 1990, 1991, 1992, 1993, 1994, 1995, 1998, 1999, 2001, | |
4 | @c 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc. | |
5 | @c See the file elisp.texi for copying conditions. | |
6 | @setfilename ../info/loading | |
7 | @node Loading, Byte Compilation, Customization, Top | |
8 | @chapter Loading | |
9 | @cindex loading | |
10 | @cindex library | |
11 | @cindex Lisp library | |
12 | ||
13 | Loading a file of Lisp code means bringing its contents into the Lisp | |
14 | environment in the form of Lisp objects. Emacs finds and opens the | |
15 | file, reads the text, evaluates each form, and then closes the file. | |
16 | ||
17 | The load functions evaluate all the expressions in a file just | |
18 | as the @code{eval-buffer} function evaluates all the | |
19 | expressions in a buffer. The difference is that the load functions | |
20 | read and evaluate the text in the file as found on disk, not the text | |
21 | in an Emacs buffer. | |
22 | ||
23 | @cindex top-level form | |
24 | The loaded file must contain Lisp expressions, either as source code | |
25 | or as byte-compiled code. Each form in the file is called a | |
26 | @dfn{top-level form}. There is no special format for the forms in a | |
27 | loadable file; any form in a file may equally well be typed directly | |
28 | into a buffer and evaluated there. (Indeed, most code is tested this | |
29 | way.) Most often, the forms are function definitions and variable | |
30 | definitions. | |
31 | ||
32 | A file containing Lisp code is often called a @dfn{library}. Thus, | |
33 | the ``Rmail library'' is a file containing code for Rmail mode. | |
34 | Similarly, a ``Lisp library directory'' is a directory of files | |
35 | containing Lisp code. | |
36 | ||
37 | @menu | |
38 | * How Programs Do Loading:: The @code{load} function and others. | |
39 | * Load Suffixes:: Details about the suffixes that @code{load} tries. | |
40 | * Library Search:: Finding a library to load. | |
41 | * Loading Non-ASCII:: Non-@acronym{ASCII} characters in Emacs Lisp files. | |
42 | * Autoload:: Setting up a function to autoload. | |
43 | * Repeated Loading:: Precautions about loading a file twice. | |
44 | * Named Features:: Loading a library if it isn't already loaded. | |
45 | * Where Defined:: Finding which file defined a certain symbol. | |
46 | * Unloading:: How to "unload" a library that was loaded. | |
47 | * Hooks for Loading:: Providing code to be run when | |
48 | particular libraries are loaded. | |
49 | @end menu | |
50 | ||
51 | @node How Programs Do Loading | |
52 | @section How Programs Do Loading | |
53 | ||
54 | Emacs Lisp has several interfaces for loading. For example, | |
55 | @code{autoload} creates a placeholder object for a function defined in a | |
56 | file; trying to call the autoloading function loads the file to get the | |
57 | function's real definition (@pxref{Autoload}). @code{require} loads a | |
58 | file if it isn't already loaded (@pxref{Named Features}). Ultimately, | |
59 | all these facilities call the @code{load} function to do the work. | |
60 | ||
61 | @defun load filename &optional missing-ok nomessage nosuffix must-suffix | |
62 | This function finds and opens a file of Lisp code, evaluates all the | |
63 | forms in it, and closes the file. | |
64 | ||
65 | To find the file, @code{load} first looks for a file named | |
66 | @file{@var{filename}.elc}, that is, for a file whose name is | |
67 | @var{filename} with the extension @samp{.elc} appended. If such a | |
68 | file exists, it is loaded. If there is no file by that name, then | |
69 | @code{load} looks for a file named @file{@var{filename}.el}. If that | |
70 | file exists, it is loaded. Finally, if neither of those names is | |
71 | found, @code{load} looks for a file named @var{filename} with nothing | |
72 | appended, and loads it if it exists. (The @code{load} function is not | |
73 | clever about looking at @var{filename}. In the perverse case of a | |
74 | file named @file{foo.el.el}, evaluation of @code{(load "foo.el")} will | |
75 | indeed find it.) | |
76 | ||
77 | If Auto Compression mode is enabled, as it is by default, then if | |
78 | @code{load} can not find a file, it searches for a compressed version | |
79 | of the file before trying other file names. It decompresses and loads | |
80 | it if it exists. It looks for compressed versions by appending each | |
81 | of the suffixes in @code{jka-compr-load-suffixes} to the file name. | |
82 | The value of this variable must be a list of strings. Its standard | |
83 | value is @code{(".gz")}. | |
84 | ||
85 | If the optional argument @var{nosuffix} is non-@code{nil}, then | |
86 | @code{load} does not try the suffixes @samp{.elc} and @samp{.el}. In | |
87 | this case, you must specify the precise file name you want, except | |
88 | that, if Auto Compression mode is enabled, @code{load} will still use | |
89 | @code{jka-compr-load-suffixes} to find compressed versions. By | |
90 | specifying the precise file name and using @code{t} for | |
91 | @var{nosuffix}, you can prevent perverse file names such as | |
92 | @file{foo.el.el} from being tried. | |
93 | ||
94 | If the optional argument @var{must-suffix} is non-@code{nil}, then | |
95 | @code{load} insists that the file name used must end in either | |
96 | @samp{.el} or @samp{.elc} (possibly extended with a compression | |
97 | suffix), unless it contains an explicit directory name. | |
98 | ||
99 | If @var{filename} is a relative file name, such as @file{foo} or | |
100 | @file{baz/foo.bar}, @code{load} searches for the file using the variable | |
101 | @code{load-path}. It appends @var{filename} to each of the directories | |
102 | listed in @code{load-path}, and loads the first file it finds whose name | |
103 | matches. The current default directory is tried only if it is specified | |
104 | in @code{load-path}, where @code{nil} stands for the default directory. | |
105 | @code{load} tries all three possible suffixes in the first directory in | |
106 | @code{load-path}, then all three suffixes in the second directory, and | |
107 | so on. @xref{Library Search}. | |
108 | ||
109 | If you get a warning that @file{foo.elc} is older than @file{foo.el}, it | |
110 | means you should consider recompiling @file{foo.el}. @xref{Byte | |
111 | Compilation}. | |
112 | ||
113 | When loading a source file (not compiled), @code{load} performs | |
114 | character set translation just as Emacs would do when visiting the file. | |
115 | @xref{Coding Systems}. | |
116 | ||
117 | Messages like @samp{Loading foo...} and @samp{Loading foo...done} appear | |
118 | in the echo area during loading unless @var{nomessage} is | |
119 | non-@code{nil}. | |
120 | ||
121 | @cindex load errors | |
122 | Any unhandled errors while loading a file terminate loading. If the | |
123 | load was done for the sake of @code{autoload}, any function definitions | |
124 | made during the loading are undone. | |
125 | ||
126 | @kindex file-error | |
127 | If @code{load} can't find the file to load, then normally it signals the | |
128 | error @code{file-error} (with @samp{Cannot open load file | |
129 | @var{filename}}). But if @var{missing-ok} is non-@code{nil}, then | |
130 | @code{load} just returns @code{nil}. | |
131 | ||
132 | You can use the variable @code{load-read-function} to specify a function | |
133 | for @code{load} to use instead of @code{read} for reading expressions. | |
134 | See below. | |
135 | ||
136 | @code{load} returns @code{t} if the file loads successfully. | |
137 | @end defun | |
138 | ||
139 | @deffn Command load-file filename | |
140 | This command loads the file @var{filename}. If @var{filename} is a | |
141 | relative file name, then the current default directory is assumed. | |
142 | This command does not use @code{load-path}, and does not append | |
143 | suffixes. However, it does look for compressed versions (if Auto | |
144 | Compression Mode is enabled). Use this command if you wish to specify | |
145 | precisely the file name to load. | |
146 | @end deffn | |
147 | ||
148 | @deffn Command load-library library | |
149 | This command loads the library named @var{library}. It is equivalent to | |
150 | @code{load}, except in how it reads its argument interactively. | |
151 | @end deffn | |
152 | ||
153 | @defvar load-in-progress | |
154 | This variable is non-@code{nil} if Emacs is in the process of loading a | |
155 | file, and it is @code{nil} otherwise. | |
156 | @end defvar | |
157 | ||
158 | @defvar load-read-function | |
159 | @anchor{Definition of load-read-function} | |
160 | @c do not allow page break at anchor; work around Texinfo deficiency. | |
161 | This variable specifies an alternate expression-reading function for | |
162 | @code{load} and @code{eval-region} to use instead of @code{read}. | |
163 | The function should accept one argument, just as @code{read} does. | |
164 | ||
165 | Normally, the variable's value is @code{nil}, which means those | |
166 | functions should use @code{read}. | |
167 | ||
168 | Instead of using this variable, it is cleaner to use another, newer | |
169 | feature: to pass the function as the @var{read-function} argument to | |
170 | @code{eval-region}. @xref{Definition of eval-region,, Eval}. | |
171 | @end defvar | |
172 | ||
173 | For information about how @code{load} is used in building Emacs, see | |
174 | @ref{Building Emacs}. | |
175 | ||
176 | @node Load Suffixes | |
177 | @section Load Suffixes | |
178 | We now describe some technical details about the exact suffixes that | |
179 | @code{load} tries. | |
180 | ||
181 | @defvar load-suffixes | |
182 | This is a list of suffixes indicating (compiled or source) Emacs Lisp | |
183 | files. It should not include the empty string. @code{load} uses | |
184 | these suffixes in order when it appends Lisp suffixes to the specified | |
185 | file name. The standard value is @code{(".elc" ".el")} which produces | |
186 | the behavior described in the previous section. | |
187 | @end defvar | |
188 | ||
189 | @defvar load-file-rep-suffixes | |
190 | This is a list of suffixes that indicate representations of the same | |
191 | file. This list should normally start with the empty string. | |
192 | When @code{load} searches for a file it appends the suffixes in this | |
193 | list, in order, to the file name, before searching for another file. | |
194 | ||
195 | Enabling Auto Compression mode appends the suffixes in | |
196 | @code{jka-compr-load-suffixes} to this list and disabling Auto | |
197 | Compression mode removes them again. The standard value of | |
198 | @code{load-file-rep-suffixes} if Auto Compression mode is disabled is | |
199 | @code{("")}. Given that the standard value of | |
200 | @code{jka-compr-load-suffixes} is @code{(".gz")}, the standard value | |
201 | of @code{load-file-rep-suffixes} if Auto Compression mode is enabled | |
202 | is @code{("" ".gz")}. | |
203 | @end defvar | |
204 | ||
205 | @defun get-load-suffixes | |
206 | This function returns the list of all suffixes that @code{load} should | |
207 | try, in order, when its @var{must-suffix} argument is non-@code{nil}. | |
208 | This takes both @code{load-suffixes} and @code{load-file-rep-suffixes} | |
209 | into account. If @code{load-suffixes}, @code{jka-compr-load-suffixes} | |
210 | and @code{load-file-rep-suffixes} all have their standard values, this | |
211 | function returns @code{(".elc" ".elc.gz" ".el" ".el.gz")} if Auto | |
212 | Compression mode is enabled and @code{(".elc" ".el")} if Auto | |
213 | Compression mode is disabled. | |
214 | @end defun | |
215 | ||
216 | To summarize, @code{load} normally first tries the suffixes in the | |
217 | value of @code{(get-load-suffixes)} and then those in | |
218 | @code{load-file-rep-suffixes}. If @var{nosuffix} is non-@code{nil}, | |
219 | it skips the former group, and if @var{must-suffix} is non-@code{nil}, | |
220 | it skips the latter group. | |
221 | ||
222 | @node Library Search | |
223 | @section Library Search | |
224 | @cindex library search | |
225 | @cindex find library | |
226 | ||
227 | When Emacs loads a Lisp library, it searches for the library | |
228 | in a list of directories specified by the variable @code{load-path}. | |
229 | ||
230 | @defopt load-path | |
231 | @cindex @code{EMACSLOADPATH} environment variable | |
232 | The value of this variable is a list of directories to search when | |
233 | loading files with @code{load}. Each element is a string (which must be | |
234 | a directory name) or @code{nil} (which stands for the current working | |
235 | directory). | |
236 | @end defopt | |
237 | ||
238 | The value of @code{load-path} is initialized from the environment | |
239 | variable @code{EMACSLOADPATH}, if that exists; otherwise its default | |
240 | value is specified in @file{emacs/src/epaths.h} when Emacs is built. | |
241 | Then the list is expanded by adding subdirectories of the directories | |
242 | in the list. | |
243 | ||
244 | The syntax of @code{EMACSLOADPATH} is the same as used for @code{PATH}; | |
245 | @samp{:} (or @samp{;}, according to the operating system) separates | |
246 | directory names, and @samp{.} is used for the current default directory. | |
247 | Here is an example of how to set your @code{EMACSLOADPATH} variable from | |
248 | a @code{csh} @file{.login} file: | |
249 | ||
250 | @smallexample | |
251 | setenv EMACSLOADPATH .:/user/bil/emacs:/usr/local/share/emacs/20.3/lisp | |
252 | @end smallexample | |
253 | ||
254 | Here is how to set it using @code{sh}: | |
255 | ||
256 | @smallexample | |
257 | export EMACSLOADPATH | |
258 | EMACSLOADPATH=.:/user/bil/emacs:/usr/local/share/emacs/20.3/lisp | |
259 | @end smallexample | |
260 | ||
261 | Here is an example of code you can place in your init file (@pxref{Init | |
262 | File}) to add several directories to the front of your default | |
263 | @code{load-path}: | |
264 | ||
265 | @smallexample | |
266 | @group | |
267 | (setq load-path | |
268 | (append (list nil "/user/bil/emacs" | |
269 | "/usr/local/lisplib" | |
270 | "~/emacs") | |
271 | load-path)) | |
272 | @end group | |
273 | @end smallexample | |
274 | ||
275 | @c Wordy to rid us of an overfull hbox. --rjc 15mar92 | |
276 | @noindent | |
277 | In this example, the path searches the current working directory first, | |
278 | followed then by the @file{/user/bil/emacs} directory, the | |
279 | @file{/usr/local/lisplib} directory, and the @file{~/emacs} directory, | |
280 | which are then followed by the standard directories for Lisp code. | |
281 | ||
282 | Dumping Emacs uses a special value of @code{load-path}. If the value of | |
283 | @code{load-path} at the end of dumping is unchanged (that is, still the | |
284 | same special value), the dumped Emacs switches to the ordinary | |
285 | @code{load-path} value when it starts up, as described above. But if | |
286 | @code{load-path} has any other value at the end of dumping, that value | |
287 | is used for execution of the dumped Emacs also. | |
288 | ||
289 | Therefore, if you want to change @code{load-path} temporarily for | |
290 | loading a few libraries in @file{site-init.el} or @file{site-load.el}, | |
291 | you should bind @code{load-path} locally with @code{let} around the | |
292 | calls to @code{load}. | |
293 | ||
294 | The default value of @code{load-path}, when running an Emacs which has | |
295 | been installed on the system, includes two special directories (and | |
296 | their subdirectories as well): | |
297 | ||
298 | @smallexample | |
299 | "/usr/local/share/emacs/@var{version}/site-lisp" | |
300 | @end smallexample | |
301 | ||
302 | @noindent | |
303 | and | |
304 | ||
305 | @smallexample | |
306 | "/usr/local/share/emacs/site-lisp" | |
307 | @end smallexample | |
308 | ||
309 | @noindent | |
310 | The first one is for locally installed packages for a particular Emacs | |
311 | version; the second is for locally installed packages meant for use with | |
312 | all installed Emacs versions. | |
313 | ||
314 | There are several reasons why a Lisp package that works well in one | |
315 | Emacs version can cause trouble in another. Sometimes packages need | |
316 | updating for incompatible changes in Emacs; sometimes they depend on | |
317 | undocumented internal Emacs data that can change without notice; | |
318 | sometimes a newer Emacs version incorporates a version of the package, | |
319 | and should be used only with that version. | |
320 | ||
321 | Emacs finds these directories' subdirectories and adds them to | |
322 | @code{load-path} when it starts up. Both immediate subdirectories and | |
323 | subdirectories multiple levels down are added to @code{load-path}. | |
324 | ||
325 | Not all subdirectories are included, though. Subdirectories whose | |
326 | names do not start with a letter or digit are excluded. Subdirectories | |
327 | named @file{RCS} or @file{CVS} are excluded. Also, a subdirectory which | |
328 | contains a file named @file{.nosearch} is excluded. You can use these | |
329 | methods to prevent certain subdirectories of the @file{site-lisp} | |
330 | directories from being searched. | |
331 | ||
332 | If you run Emacs from the directory where it was built---that is, an | |
333 | executable that has not been formally installed---then @code{load-path} | |
334 | normally contains two additional directories. These are the @code{lisp} | |
335 | and @code{site-lisp} subdirectories of the main build directory. (Both | |
336 | are represented as absolute file names.) | |
337 | ||
338 | @deffn Command locate-library library &optional nosuffix path interactive-call | |
339 | This command finds the precise file name for library @var{library}. It | |
340 | searches for the library in the same way @code{load} does, and the | |
341 | argument @var{nosuffix} has the same meaning as in @code{load}: don't | |
342 | add suffixes @samp{.elc} or @samp{.el} to the specified name | |
343 | @var{library}. | |
344 | ||
345 | If the @var{path} is non-@code{nil}, that list of directories is used | |
346 | instead of @code{load-path}. | |
347 | ||
348 | When @code{locate-library} is called from a program, it returns the file | |
349 | name as a string. When the user runs @code{locate-library} | |
350 | interactively, the argument @var{interactive-call} is @code{t}, and this | |
351 | tells @code{locate-library} to display the file name in the echo area. | |
352 | @end deffn | |
353 | ||
354 | @node Loading Non-ASCII | |
355 | @section Loading Non-@acronym{ASCII} Characters | |
356 | ||
357 | When Emacs Lisp programs contain string constants with non-@acronym{ASCII} | |
358 | characters, these can be represented within Emacs either as unibyte | |
359 | strings or as multibyte strings (@pxref{Text Representations}). Which | |
360 | representation is used depends on how the file is read into Emacs. If | |
361 | it is read with decoding into multibyte representation, the text of the | |
362 | Lisp program will be multibyte text, and its string constants will be | |
363 | multibyte strings. If a file containing Latin-1 characters (for | |
364 | example) is read without decoding, the text of the program will be | |
365 | unibyte text, and its string constants will be unibyte strings. | |
366 | @xref{Coding Systems}. | |
367 | ||
368 | To make the results more predictable, Emacs always performs decoding | |
369 | into the multibyte representation when loading Lisp files, even if it | |
370 | was started with the @samp{--unibyte} option. This means that string | |
371 | constants with non-@acronym{ASCII} characters translate into multibyte | |
372 | strings. The only exception is when a particular file specifies no | |
373 | decoding. | |
374 | ||
375 | The reason Emacs is designed this way is so that Lisp programs give | |
376 | predictable results, regardless of how Emacs was started. In addition, | |
377 | this enables programs that depend on using multibyte text to work even | |
378 | in a unibyte Emacs. Of course, such programs should be designed to | |
379 | notice whether the user prefers unibyte or multibyte text, by checking | |
380 | @code{default-enable-multibyte-characters}, and convert representations | |
381 | appropriately. | |
382 | ||
383 | In most Emacs Lisp programs, the fact that non-@acronym{ASCII} strings are | |
384 | multibyte strings should not be noticeable, since inserting them in | |
385 | unibyte buffers converts them to unibyte automatically. However, if | |
386 | this does make a difference, you can force a particular Lisp file to be | |
387 | interpreted as unibyte by writing @samp{-*-unibyte: t;-*-} in a | |
388 | comment on the file's first line. With that designator, the file will | |
389 | unconditionally be interpreted as unibyte, even in an ordinary | |
390 | multibyte Emacs session. This can matter when making keybindings to | |
391 | non-@acronym{ASCII} characters written as @code{?v@var{literal}}. | |
392 | ||
393 | @node Autoload | |
394 | @section Autoload | |
395 | @cindex autoload | |
396 | ||
397 | The @dfn{autoload} facility allows you to make a function or macro | |
398 | known in Lisp, but put off loading the file that defines it. The first | |
399 | call to the function automatically reads the proper file to install the | |
400 | real definition and other associated code, then runs the real definition | |
401 | as if it had been loaded all along. | |
402 | ||
403 | There are two ways to set up an autoloaded function: by calling | |
404 | @code{autoload}, and by writing a special ``magic'' comment in the | |
405 | source before the real definition. @code{autoload} is the low-level | |
406 | primitive for autoloading; any Lisp program can call @code{autoload} at | |
407 | any time. Magic comments are the most convenient way to make a function | |
408 | autoload, for packages installed along with Emacs. These comments do | |
409 | nothing on their own, but they serve as a guide for the command | |
410 | @code{update-file-autoloads}, which constructs calls to @code{autoload} | |
411 | and arranges to execute them when Emacs is built. | |
412 | ||
413 | @defun autoload function filename &optional docstring interactive type | |
414 | This function defines the function (or macro) named @var{function} so as | |
415 | to load automatically from @var{filename}. The string @var{filename} | |
416 | specifies the file to load to get the real definition of @var{function}. | |
417 | ||
418 | If @var{filename} does not contain either a directory name, or the | |
419 | suffix @code{.el} or @code{.elc}, then @code{autoload} insists on adding | |
420 | one of these suffixes, and it will not load from a file whose name is | |
421 | just @var{filename} with no added suffix. (The variable | |
422 | @code{load-suffixes} specifies the exact required suffixes.) | |
423 | ||
424 | The argument @var{docstring} is the documentation string for the | |
425 | function. Specifying the documentation string in the call to | |
426 | @code{autoload} makes it possible to look at the documentation without | |
427 | loading the function's real definition. Normally, this should be | |
428 | identical to the documentation string in the function definition | |
429 | itself. If it isn't, the function definition's documentation string | |
430 | takes effect when it is loaded. | |
431 | ||
432 | If @var{interactive} is non-@code{nil}, that says @var{function} can be | |
433 | called interactively. This lets completion in @kbd{M-x} work without | |
434 | loading @var{function}'s real definition. The complete interactive | |
435 | specification is not given here; it's not needed unless the user | |
436 | actually calls @var{function}, and when that happens, it's time to load | |
437 | the real definition. | |
438 | ||
439 | You can autoload macros and keymaps as well as ordinary functions. | |
440 | Specify @var{type} as @code{macro} if @var{function} is really a macro. | |
441 | Specify @var{type} as @code{keymap} if @var{function} is really a | |
442 | keymap. Various parts of Emacs need to know this information without | |
443 | loading the real definition. | |
444 | ||
445 | An autoloaded keymap loads automatically during key lookup when a prefix | |
446 | key's binding is the symbol @var{function}. Autoloading does not occur | |
447 | for other kinds of access to the keymap. In particular, it does not | |
448 | happen when a Lisp program gets the keymap from the value of a variable | |
449 | and calls @code{define-key}; not even if the variable name is the same | |
450 | symbol @var{function}. | |
451 | ||
452 | @cindex function cell in autoload | |
453 | If @var{function} already has a non-void function definition that is not | |
454 | an autoload object, @code{autoload} does nothing and returns @code{nil}. | |
455 | If the function cell of @var{function} is void, or is already an autoload | |
456 | object, then it is defined as an autoload object like this: | |
457 | ||
458 | @example | |
459 | (autoload @var{filename} @var{docstring} @var{interactive} @var{type}) | |
460 | @end example | |
461 | ||
462 | For example, | |
463 | ||
464 | @example | |
465 | @group | |
466 | (symbol-function 'run-prolog) | |
467 | @result{} (autoload "prolog" 169681 t nil) | |
468 | @end group | |
469 | @end example | |
470 | ||
471 | @noindent | |
472 | In this case, @code{"prolog"} is the name of the file to load, 169681 | |
473 | refers to the documentation string in the | |
474 | @file{emacs/etc/DOC-@var{version}} file (@pxref{Documentation Basics}), | |
475 | @code{t} means the function is interactive, and @code{nil} that it is | |
476 | not a macro or a keymap. | |
477 | @end defun | |
478 | ||
479 | @cindex autoload errors | |
480 | The autoloaded file usually contains other definitions and may require | |
481 | or provide one or more features. If the file is not completely loaded | |
482 | (due to an error in the evaluation of its contents), any function | |
483 | definitions or @code{provide} calls that occurred during the load are | |
484 | undone. This is to ensure that the next attempt to call any function | |
485 | autoloading from this file will try again to load the file. If not for | |
486 | this, then some of the functions in the file might be defined by the | |
487 | aborted load, but fail to work properly for the lack of certain | |
488 | subroutines not loaded successfully because they come later in the file. | |
489 | ||
490 | If the autoloaded file fails to define the desired Lisp function or | |
491 | macro, then an error is signaled with data @code{"Autoloading failed to | |
492 | define function @var{function-name}"}. | |
493 | ||
494 | @findex update-file-autoloads | |
495 | @findex update-directory-autoloads | |
496 | @cindex magic autoload comment | |
497 | @cindex autoload cookie | |
498 | @anchor{autoload cookie} | |
499 | A magic autoload comment (often called an @dfn{autoload cookie}) | |
500 | consists of @samp{;;;###autoload}, on a line by itself, | |
501 | just before the real definition of the function in its | |
502 | autoloadable source file. The command @kbd{M-x update-file-autoloads} | |
503 | writes a corresponding @code{autoload} call into @file{loaddefs.el}. | |
504 | Building Emacs loads @file{loaddefs.el} and thus calls @code{autoload}. | |
505 | @kbd{M-x update-directory-autoloads} is even more powerful; it updates | |
506 | autoloads for all files in the current directory. | |
507 | ||
508 | The same magic comment can copy any kind of form into | |
509 | @file{loaddefs.el}. If the form following the magic comment is not a | |
510 | function-defining form or a @code{defcustom} form, it is copied | |
511 | verbatim. ``Function-defining forms'' include @code{define-skeleton}, | |
512 | @code{define-derived-mode}, @code{define-generic-mode} and | |
513 | @code{define-minor-mode} as well as @code{defun} and | |
514 | @code{defmacro}. To save space, a @code{defcustom} form is converted to | |
515 | a @code{defvar} in @file{loaddefs.el}, with some additional information | |
516 | if it uses @code{:require}. | |
517 | ||
518 | You can also use a magic comment to execute a form at build time | |
519 | @emph{without} executing it when the file itself is loaded. To do this, | |
520 | write the form @emph{on the same line} as the magic comment. Since it | |
521 | is in a comment, it does nothing when you load the source file; but | |
522 | @kbd{M-x update-file-autoloads} copies it to @file{loaddefs.el}, where | |
523 | it is executed while building Emacs. | |
524 | ||
525 | The following example shows how @code{doctor} is prepared for | |
526 | autoloading with a magic comment: | |
527 | ||
528 | @smallexample | |
529 | ;;;###autoload | |
530 | (defun doctor () | |
531 | "Switch to *doctor* buffer and start giving psychotherapy." | |
532 | (interactive) | |
533 | (switch-to-buffer "*doctor*") | |
534 | (doctor-mode)) | |
535 | @end smallexample | |
536 | ||
537 | @noindent | |
538 | Here's what that produces in @file{loaddefs.el}: | |
539 | ||
540 | @smallexample | |
541 | (autoload (quote doctor) "doctor" "\ | |
542 | Switch to *doctor* buffer and start giving psychotherapy. | |
543 | ||
544 | \(fn)" t nil) | |
545 | @end smallexample | |
546 | ||
547 | @noindent | |
548 | @cindex @code{fn} in function's documentation string | |
549 | The backslash and newline immediately following the double-quote are a | |
550 | convention used only in the preloaded uncompiled Lisp files such as | |
551 | @file{loaddefs.el}; they tell @code{make-docfile} to put the | |
552 | documentation string in the @file{etc/DOC} file. @xref{Building Emacs}. | |
553 | See also the commentary in @file{lib-src/make-docfile.c}. @samp{(fn)} | |
554 | in the usage part of the documentation string is replaced with the | |
555 | function's name when the various help functions (@pxref{Help | |
556 | Functions}) display it. | |
557 | ||
558 | If you write a function definition with an unusual macro that is not | |
559 | one of the known and recognized function definition methods, use of an | |
560 | ordinary magic autoload comment would copy the whole definition into | |
561 | @code{loaddefs.el}. That is not desirable. You can put the desired | |
562 | @code{autoload} call into @code{loaddefs.el} instead by writing this: | |
563 | ||
564 | @smallexample | |
565 | ;;;###autoload (autoload 'foo "myfile") | |
566 | (mydefunmacro foo | |
567 | ...) | |
568 | @end smallexample | |
569 | ||
570 | @node Repeated Loading | |
571 | @section Repeated Loading | |
572 | @cindex repeated loading | |
573 | ||
574 | You can load a given file more than once in an Emacs session. For | |
575 | example, after you have rewritten and reinstalled a function definition | |
576 | by editing it in a buffer, you may wish to return to the original | |
577 | version; you can do this by reloading the file it came from. | |
578 | ||
579 | When you load or reload files, bear in mind that the @code{load} and | |
580 | @code{load-library} functions automatically load a byte-compiled file | |
581 | rather than a non-compiled file of similar name. If you rewrite a file | |
582 | that you intend to save and reinstall, you need to byte-compile the new | |
583 | version; otherwise Emacs will load the older, byte-compiled file instead | |
584 | of your newer, non-compiled file! If that happens, the message | |
585 | displayed when loading the file includes, @samp{(compiled; note, source is | |
586 | newer)}, to remind you to recompile it. | |
587 | ||
588 | When writing the forms in a Lisp library file, keep in mind that the | |
589 | file might be loaded more than once. For example, think about whether | |
590 | each variable should be reinitialized when you reload the library; | |
591 | @code{defvar} does not change the value if the variable is already | |
592 | initialized. (@xref{Defining Variables}.) | |
593 | ||
594 | The simplest way to add an element to an alist is like this: | |
595 | ||
596 | @example | |
597 | (push '(leif-mode " Leif") minor-mode-alist) | |
598 | @end example | |
599 | ||
600 | @noindent | |
601 | But this would add multiple elements if the library is reloaded. | |
602 | To avoid the problem, write this: | |
603 | ||
604 | @example | |
605 | (or (assq 'leif-mode minor-mode-alist) | |
606 | (push '(leif-mode " Leif") minor-mode-alist)) | |
607 | @end example | |
608 | ||
609 | @noindent | |
610 | or this: | |
611 | ||
612 | @example | |
613 | (add-to-list '(leif-mode " Leif") minor-mode-alist) | |
614 | @end example | |
615 | ||
616 | Occasionally you will want to test explicitly whether a library has | |
617 | already been loaded. Here's one way to test, in a library, whether it | |
618 | has been loaded before: | |
619 | ||
620 | @example | |
621 | (defvar foo-was-loaded nil) | |
622 | ||
623 | (unless foo-was-loaded | |
624 | @var{execute-first-time-only} | |
625 | (setq foo-was-loaded t)) | |
626 | @end example | |
627 | ||
628 | @noindent | |
629 | If the library uses @code{provide} to provide a named feature, you can | |
630 | use @code{featurep} earlier in the file to test whether the | |
631 | @code{provide} call has been executed before. | |
632 | @ifnottex | |
633 | @xref{Named Features}. | |
634 | @end ifnottex | |
635 | ||
636 | @node Named Features | |
637 | @section Features | |
638 | @cindex features | |
639 | @cindex requiring features | |
640 | @cindex providing features | |
641 | ||
642 | @code{provide} and @code{require} are an alternative to | |
643 | @code{autoload} for loading files automatically. They work in terms of | |
644 | named @dfn{features}. Autoloading is triggered by calling a specific | |
645 | function, but a feature is loaded the first time another program asks | |
646 | for it by name. | |
647 | ||
648 | A feature name is a symbol that stands for a collection of functions, | |
649 | variables, etc. The file that defines them should @dfn{provide} the | |
650 | feature. Another program that uses them may ensure they are defined by | |
651 | @dfn{requiring} the feature. This loads the file of definitions if it | |
652 | hasn't been loaded already. | |
653 | ||
654 | To require the presence of a feature, call @code{require} with the | |
655 | feature name as argument. @code{require} looks in the global variable | |
656 | @code{features} to see whether the desired feature has been provided | |
657 | already. If not, it loads the feature from the appropriate file. This | |
658 | file should call @code{provide} at the top level to add the feature to | |
659 | @code{features}; if it fails to do so, @code{require} signals an error. | |
660 | @cindex load error with require | |
661 | ||
662 | For example, in @file{emacs/lisp/prolog.el}, | |
663 | the definition for @code{run-prolog} includes the following code: | |
664 | ||
665 | @smallexample | |
666 | (defun run-prolog () | |
667 | "Run an inferior Prolog process, with I/O via buffer *prolog*." | |
668 | (interactive) | |
669 | (require 'comint) | |
670 | (switch-to-buffer (make-comint "prolog" prolog-program-name)) | |
671 | (inferior-prolog-mode)) | |
672 | @end smallexample | |
673 | ||
674 | @noindent | |
675 | The expression @code{(require 'comint)} loads the file @file{comint.el} | |
676 | if it has not yet been loaded. This ensures that @code{make-comint} is | |
677 | defined. Features are normally named after the files that provide them, | |
678 | so that @code{require} need not be given the file name. | |
679 | ||
680 | The @file{comint.el} file contains the following top-level expression: | |
681 | ||
682 | @smallexample | |
683 | (provide 'comint) | |
684 | @end smallexample | |
685 | ||
686 | @noindent | |
687 | This adds @code{comint} to the global @code{features} list, so that | |
688 | @code{(require 'comint)} will henceforth know that nothing needs to be | |
689 | done. | |
690 | ||
691 | @cindex byte-compiling @code{require} | |
692 | When @code{require} is used at top level in a file, it takes effect | |
693 | when you byte-compile that file (@pxref{Byte Compilation}) as well as | |
694 | when you load it. This is in case the required package contains macros | |
695 | that the byte compiler must know about. It also avoids byte-compiler | |
696 | warnings for functions and variables defined in the file loaded with | |
697 | @code{require}. | |
698 | ||
699 | Although top-level calls to @code{require} are evaluated during | |
700 | byte compilation, @code{provide} calls are not. Therefore, you can | |
701 | ensure that a file of definitions is loaded before it is byte-compiled | |
702 | by including a @code{provide} followed by a @code{require} for the same | |
703 | feature, as in the following example. | |
704 | ||
705 | @smallexample | |
706 | @group | |
707 | (provide 'my-feature) ; @r{Ignored by byte compiler,} | |
708 | ; @r{evaluated by @code{load}.} | |
709 | (require 'my-feature) ; @r{Evaluated by byte compiler.} | |
710 | @end group | |
711 | @end smallexample | |
712 | ||
713 | @noindent | |
714 | The compiler ignores the @code{provide}, then processes the | |
715 | @code{require} by loading the file in question. Loading the file does | |
716 | execute the @code{provide} call, so the subsequent @code{require} call | |
717 | does nothing when the file is loaded. | |
718 | ||
719 | @defun provide feature &optional subfeatures | |
720 | This function announces that @var{feature} is now loaded, or being | |
721 | loaded, into the current Emacs session. This means that the facilities | |
722 | associated with @var{feature} are or will be available for other Lisp | |
723 | programs. | |
724 | ||
725 | The direct effect of calling @code{provide} is to add @var{feature} to | |
726 | the front of the list @code{features} if it is not already in the list. | |
727 | The argument @var{feature} must be a symbol. @code{provide} returns | |
728 | @var{feature}. | |
729 | ||
730 | If provided, @var{subfeatures} should be a list of symbols indicating | |
731 | a set of specific subfeatures provided by this version of | |
732 | @var{feature}. You can test the presence of a subfeature using | |
733 | @code{featurep}. The idea of subfeatures is that you use them when a | |
734 | package (which is one @var{feature}) is complex enough to make it | |
735 | useful to give names to various parts or functionalities of the | |
736 | package, which might or might not be loaded, or might or might not be | |
737 | present in a given version. @xref{Network Feature Testing}, for | |
738 | an example. | |
739 | ||
740 | @smallexample | |
741 | features | |
742 | @result{} (bar bish) | |
743 | ||
744 | (provide 'foo) | |
745 | @result{} foo | |
746 | features | |
747 | @result{} (foo bar bish) | |
748 | @end smallexample | |
749 | ||
750 | When a file is loaded to satisfy an autoload, and it stops due to an | |
751 | error in the evaluation of its contents, any function definitions or | |
752 | @code{provide} calls that occurred during the load are undone. | |
753 | @xref{Autoload}. | |
754 | @end defun | |
755 | ||
756 | @defun require feature &optional filename noerror | |
757 | This function checks whether @var{feature} is present in the current | |
758 | Emacs session (using @code{(featurep @var{feature})}; see below). The | |
759 | argument @var{feature} must be a symbol. | |
760 | ||
761 | If the feature is not present, then @code{require} loads @var{filename} | |
762 | with @code{load}. If @var{filename} is not supplied, then the name of | |
763 | the symbol @var{feature} is used as the base file name to load. | |
764 | However, in this case, @code{require} insists on finding @var{feature} | |
765 | with an added @samp{.el} or @samp{.elc} suffix (possibly extended with | |
766 | a compression suffix); a file whose name is just @var{feature} won't | |
767 | be used. (The variable @code{load-suffixes} specifies the exact | |
768 | required Lisp suffixes.) | |
769 | ||
770 | If @var{noerror} is non-@code{nil}, that suppresses errors from actual | |
771 | loading of the file. In that case, @code{require} returns @code{nil} | |
772 | if loading the file fails. Normally, @code{require} returns | |
773 | @var{feature}. | |
774 | ||
775 | If loading the file succeeds but does not provide @var{feature}, | |
776 | @code{require} signals an error, @samp{Required feature @var{feature} | |
777 | was not provided}. | |
778 | @end defun | |
779 | ||
780 | @defun featurep feature &optional subfeature | |
781 | This function returns @code{t} if @var{feature} has been provided in | |
782 | the current Emacs session (i.e.@:, if @var{feature} is a member of | |
783 | @code{features}.) If @var{subfeature} is non-@code{nil}, then the | |
784 | function returns @code{t} only if that subfeature is provided as well | |
785 | (i.e.@: if @var{subfeature} is a member of the @code{subfeature} | |
786 | property of the @var{feature} symbol.) | |
787 | @end defun | |
788 | ||
789 | @defvar features | |
790 | The value of this variable is a list of symbols that are the features | |
791 | loaded in the current Emacs session. Each symbol was put in this list | |
792 | with a call to @code{provide}. The order of the elements in the | |
793 | @code{features} list is not significant. | |
794 | @end defvar | |
795 | ||
796 | @node Where Defined | |
797 | @section Which File Defined a Certain Symbol | |
798 | ||
799 | @defun symbol-file symbol &optional type | |
800 | This function returns the name of the file that defined @var{symbol}. | |
801 | If @var{type} is @code{nil}, then any kind of definition is | |
802 | acceptable. If @var{type} is @code{defun} or @code{defvar}, that | |
803 | specifies function definition only or variable definition only. | |
804 | ||
805 | The value is normally an absolute file name. It can also be | |
806 | @code{nil}, if the definition is not associated with any file. | |
807 | @end defun | |
808 | ||
809 | The basis for @code{symbol-file} is the data in the variable | |
810 | @code{load-history}. | |
811 | ||
812 | @defvar load-history | |
813 | This variable's value is an alist connecting library file names with the | |
814 | names of functions and variables they define, the features they provide, | |
815 | and the features they require. | |
816 | ||
817 | Each element is a list and describes one library. The @sc{car} of the | |
818 | list is the absolute file name of the library, as a string. The rest | |
819 | of the list elements have these forms: | |
820 | ||
821 | @table @code | |
822 | @item @var{var} | |
823 | The symbol @var{var} was defined as a variable. | |
824 | @item (defun . @var{fun}) | |
825 | The function @var{fun} was defined. | |
826 | @item (t . @var{fun}) | |
827 | The function @var{fun} was previously an autoload before this library | |
828 | redefined it as a function. The following element is always | |
829 | @code{(defun . @var{fun})}, which represents defining @var{fun} as a | |
830 | function. | |
831 | @item (autoload . @var{fun}) | |
832 | The function @var{fun} was defined as an autoload. | |
833 | @item (require . @var{feature}) | |
834 | The feature @var{feature} was required. | |
835 | @item (provide . @var{feature}) | |
836 | The feature @var{feature} was provided. | |
837 | @end table | |
838 | ||
839 | The value of @code{load-history} may have one element whose @sc{car} is | |
840 | @code{nil}. This element describes definitions made with | |
841 | @code{eval-buffer} on a buffer that is not visiting a file. | |
842 | @end defvar | |
843 | ||
844 | The command @code{eval-region} updates @code{load-history}, but does so | |
845 | by adding the symbols defined to the element for the file being visited, | |
846 | rather than replacing that element. @xref{Eval}. | |
847 | ||
848 | @node Unloading | |
849 | @section Unloading | |
850 | @cindex unloading packages | |
851 | ||
852 | @c Emacs 19 feature | |
853 | You can discard the functions and variables loaded by a library to | |
854 | reclaim memory for other Lisp objects. To do this, use the function | |
855 | @code{unload-feature}: | |
856 | ||
857 | @deffn Command unload-feature feature &optional force | |
858 | This command unloads the library that provided feature @var{feature}. | |
859 | It undefines all functions, macros, and variables defined in that | |
860 | library with @code{defun}, @code{defalias}, @code{defsubst}, | |
861 | @code{defmacro}, @code{defconst}, @code{defvar}, and @code{defcustom}. | |
862 | It then restores any autoloads formerly associated with those symbols. | |
863 | (Loading saves these in the @code{autoload} property of the symbol.) | |
864 | ||
865 | @vindex unload-feature-special-hooks | |
866 | Before restoring the previous definitions, @code{unload-feature} runs | |
867 | @code{remove-hook} to remove functions in the library from certain | |
868 | hooks. These hooks include variables whose names end in @samp{hook} | |
869 | or @samp{-hooks}, plus those listed in | |
870 | @code{unload-feature-special-hooks}. This is to prevent Emacs from | |
871 | ceasing to function because important hooks refer to functions that | |
872 | are no longer defined. | |
873 | ||
874 | @vindex @var{feature}-unload-hook | |
875 | If these measures are not sufficient to prevent malfunction, a library | |
876 | can define an explicit unload hook. If @code{@var{feature}-unload-hook} | |
877 | is defined, it is run as a normal hook before restoring the previous | |
878 | definitions, @emph{instead of} the usual hook-removing actions. The | |
879 | unload hook ought to undo all the global state changes made by the | |
880 | library that might cease to work once the library is unloaded. | |
881 | @code{unload-feature} can cause problems with libraries that fail to do | |
882 | this, so it should be used with caution. | |
883 | ||
884 | Ordinarily, @code{unload-feature} refuses to unload a library on which | |
885 | other loaded libraries depend. (A library @var{a} depends on library | |
886 | @var{b} if @var{a} contains a @code{require} for @var{b}.) If the | |
887 | optional argument @var{force} is non-@code{nil}, dependencies are | |
888 | ignored and you can unload any library. | |
889 | @end deffn | |
890 | ||
891 | The @code{unload-feature} function is written in Lisp; its actions are | |
892 | based on the variable @code{load-history}. | |
893 | ||
894 | @defvar unload-feature-special-hooks | |
895 | This variable holds a list of hooks to be scanned before unloading a | |
896 | library, to remove functions defined in the library. | |
897 | @end defvar | |
898 | ||
899 | @node Hooks for Loading | |
900 | @section Hooks for Loading | |
901 | @cindex loading hooks | |
902 | @cindex hooks for loading | |
903 | ||
904 | You can ask for code to be executed if and when a particular library is | |
905 | loaded, by calling @code{eval-after-load}. | |
906 | ||
907 | @defun eval-after-load library form | |
908 | This function arranges to evaluate @var{form} at the end of loading | |
909 | the file @var{library}, each time @var{library} is loaded. If | |
910 | @var{library} is already loaded, it evaluates @var{form} right away. | |
911 | Don't forget to quote @var{form}! | |
912 | ||
913 | You don't need to give a directory or extension in the file name | |
914 | @var{library}---normally you just give a bare file name, like this: | |
915 | ||
916 | @example | |
917 | (eval-after-load "edebug" '(def-edebug-spec c-point t)) | |
918 | @end example | |
919 | ||
920 | To restrict which files can trigger the evaluation, include a | |
921 | directory or an extension or both in @var{library}. Only a file whose | |
922 | absolute true name (i.e., the name with all symbolic links chased out) | |
923 | matches all the given name components will match. In the following | |
924 | example, @file{my_inst.elc} or @file{my_inst.elc.gz} in some directory | |
925 | @code{..../foo/bar} will trigger the evaluation, but not | |
926 | @file{my_inst.el}: | |
927 | ||
928 | @example | |
929 | (eval-after-load "foo/bar/my_inst.elc" @dots{}) | |
930 | @end example | |
931 | ||
932 | @var{library} can also be a feature (i.e.@: a symbol), in which case | |
933 | @var{form} is evaluated when @code{(provide @var{library})} is called. | |
934 | ||
935 | An error in @var{form} does not undo the load, but does prevent | |
936 | execution of the rest of @var{form}. | |
937 | @end defun | |
938 | ||
939 | In general, well-designed Lisp programs should not use this feature. | |
940 | The clean and modular ways to interact with a Lisp library are (1) | |
941 | examine and set the library's variables (those which are meant for | |
942 | outside use), and (2) call the library's functions. If you wish to | |
943 | do (1), you can do it immediately---there is no need to wait for when | |
944 | the library is loaded. To do (2), you must load the library (preferably | |
945 | with @code{require}). | |
946 | ||
947 | But it is OK to use @code{eval-after-load} in your personal | |
948 | customizations if you don't feel they must meet the design standards for | |
949 | programs meant for wider use. | |
950 | ||
951 | @defvar after-load-alist | |
952 | This variable, an alist built by @code{eval-after-load}, holds the | |
953 | expressions to evaluate when particular libraries are loaded. Each | |
954 | element looks like this: | |
955 | ||
956 | @example | |
957 | (@var{regexp-or-feature} @var{forms}@dots{}) | |
958 | @end example | |
959 | ||
960 | The key @var{regexp-or-feature} is either a regular expression or a | |
961 | symbol, and the value is a list of forms. The forms are evaluated when | |
962 | the key matches the absolute true name of the file being | |
963 | @code{load}ed or the symbol being @code{provide}d. | |
964 | @end defvar | |
965 | ||
966 | @ignore | |
967 | arch-tag: df731f89-0900-4389-a436-9105241b6f7a | |
968 | @end ignore |