Spelling and wording fixes.

[bpt/emacs.git] / doc / lispref / display.texi

@c -*-texinfo-*-
@c This is part of the GNU Emacs Lisp Reference Manual.
@c Copyright (C) 1990-1995, 1998-2014 Free Software Foundation, Inc.
@c See the file elisp.texi for copying conditions.
@node Display
@chapter Emacs Display

  This chapter describes a number of features related to the display
that Emacs presents to the user.

@menu
* Refresh Screen::      Clearing the screen and redrawing everything on it.
* Forcing Redisplay::   Forcing redisplay.
* Truncation::          Folding or wrapping long text lines.
* The Echo Area::       Displaying messages at the bottom of the screen.
* Warnings::            Displaying warning messages for the user.
* Invisible Text::      Hiding part of the buffer text.
* Selective Display::   Hiding part of the buffer text (the old way).
* Temporary Displays::  Displays that go away automatically.
* Overlays::            Use overlays to highlight parts of the buffer.
* Width::               How wide a character or string is on the screen.
* Line Height::         Controlling the height of lines.
* Faces::               A face defines a graphics style for text characters:
                          font, colors, etc.
* Fringes::             Controlling window fringes.
* Scroll Bars::         Controlling vertical scroll bars.
* Window Dividers::     Separating windows visually.
* Display Property::    Enabling special display features.
* Images::              Displaying images in Emacs buffers.
* Buttons::             Adding clickable buttons to Emacs buffers.
* Abstract Display::    Emacs's Widget for Object Collections.
* Blinking::            How Emacs shows the matching open parenthesis.
* Character Display::   How Emacs displays individual characters.
* Beeping::             Audible signal to the user.
* Window Systems::      Which window system is being used.
* Bidirectional Display:: Display of bidirectional scripts, such as
                             Arabic and Farsi.
@end menu

@node Refresh Screen
@section Refreshing the Screen

  The function @code{redraw-frame} clears and redisplays the entire
contents of a given frame (@pxref{Frames}).  This is useful if the
screen is corrupted.

@defun redraw-frame frame
This function clears and redisplays frame @var{frame}.
@end defun

  Even more powerful is @code{redraw-display}:

@deffn Command redraw-display
This function clears and redisplays all visible frames.
@end deffn

  In Emacs, processing user input takes priority over redisplay.  If
you call these functions when input is available, they don't redisplay
immediately, but the requested redisplay does happen
eventually---after all the input has been processed.

  On text terminals, suspending and resuming Emacs normally also
refreshes the screen.  Some terminal emulators record separate
contents for display-oriented programs such as Emacs and for ordinary
sequential display.  If you are using such a terminal, you might want
to inhibit the redisplay on resumption.

@defopt no-redraw-on-reenter
@cindex suspend (cf. @code{no-redraw-on-reenter})
@cindex resume (cf. @code{no-redraw-on-reenter})
This variable controls whether Emacs redraws the entire screen after it
has been suspended and resumed.  Non-@code{nil} means there is no need
to redraw, @code{nil} means redrawing is needed.  The default is @code{nil}.
@end defopt

@node Forcing Redisplay
@section Forcing Redisplay
@cindex forcing redisplay

  Emacs normally tries to redisplay the screen whenever it waits for
input.  With the following function, you can request an immediate
attempt to redisplay, in the middle of Lisp code, without actually
waiting for input.

@defun redisplay &optional force
This function tries immediately to redisplay.  The optional argument
@var{force}, if non-@code{nil}, forces the redisplay to be performed,
instead of being preempted, even if input is pending and the variable
@code{redisplay-dont-pause} is @code{nil} (see below).  If
@code{redisplay-dont-pause} is non-@code{nil} (the default), this
function redisplays in any case, i.e., @var{force} does nothing.

The function returns @code{t} if it actually tried to redisplay, and
@code{nil} otherwise.  A value of @code{t} does not mean that
redisplay proceeded to completion; it could have been preempted by
newly arriving input.
@end defun

@defvar redisplay-dont-pause
If this variable is @code{nil}, arriving input events preempt
redisplay; Emacs avoids starting a redisplay, and stops any redisplay
that is in progress, until the input has been processed.  In
particular, @code{(redisplay)} returns @code{nil} without actually
redisplaying, if there is pending input.

The default value is @code{t}, which means that pending input does not
preempt redisplay.
@end defvar

@defvar redisplay-preemption-period
If @code{redisplay-dont-pause} is @code{nil}, this variable specifies
how many seconds Emacs waits between checks for new input during
redisplay; if input arrives during this interval, redisplay stops and
the input is processed.  The default value is 0.1; if the value is
@code{nil}, Emacs does not check for input during redisplay.

This variable has no effect when @code{redisplay-dont-pause} is
non-@code{nil} (the default).
@end defvar

@defvar pre-redisplay-function
A function run just before redisplay.  It is called with one argument,
the set of windows to redisplay.
@end defvar

  Although @code{redisplay} tries immediately to redisplay, it does
not change how Emacs decides which parts of its frame(s) to redisplay.
By contrast, the following function adds certain windows to the
pending redisplay work (as if their contents had completely changed),
but does not immediately try to perform redisplay.

@defun force-window-update &optional object
This function forces some or all windows to be updated the next time
Emacs does a redisplay.  If @var{object} is a window, that window is
to be updated.  If @var{object} is a buffer or buffer name, all
windows displaying that buffer are to be updated.  If @var{object} is
@code{nil} (or omitted), all windows are to be updated.

This function does not do a redisplay immediately; Emacs does that as
it waits for input, or when the function @code{redisplay} is called.
@end defun

@node Truncation
@section Truncation
@cindex line wrapping
@cindex line truncation
@cindex continuation lines
@cindex @samp{$} in display
@cindex @samp{\} in display

  When a line of text extends beyond the right edge of a window, Emacs
can @dfn{continue} the line (make it ``wrap'' to the next screen
line), or @dfn{truncate} the line (limit it to one screen line).  The
additional screen lines used to display a long text line are called
@dfn{continuation} lines.  Continuation is not the same as filling;
continuation happens on the screen only, not in the buffer contents,
and it breaks a line precisely at the right margin, not at a word
boundary.  @xref{Filling}.

   On a graphical display, tiny arrow images in the window fringes
indicate truncated and continued lines (@pxref{Fringes}).  On a text
terminal, a @samp{$} in the rightmost column of the window indicates
truncation; a @samp{\} on the rightmost column indicates a line that
``wraps''.  (The display table can specify alternate characters to use
for this; @pxref{Display Tables}).

@defopt truncate-lines
If this buffer-local variable is non-@code{nil}, lines that extend
beyond the right edge of the window are truncated; otherwise, they are
continued.  As a special exception, the variable
@code{truncate-partial-width-windows} takes precedence in
@dfn{partial-width} windows (i.e., windows that do not occupy the
entire frame width).
@end defopt

@defopt truncate-partial-width-windows
@cindex partial-width windows
This variable controls line truncation in @dfn{partial-width} windows.
A partial-width window is one that does not occupy the entire frame
width (@pxref{Splitting Windows}).  If the value is @code{nil}, line
truncation is determined by the variable @code{truncate-lines} (see
above).  If the value is an integer @var{n}, lines are truncated if
the partial-width window has fewer than @var{n} columns, regardless of
the value of @code{truncate-lines}; if the partial-width window has
@var{n} or more columns, line truncation is determined by
@code{truncate-lines}.  For any other non-@code{nil} value, lines are
truncated in every partial-width window, regardless of the value of
@code{truncate-lines}.
@end defopt

  When horizontal scrolling (@pxref{Horizontal Scrolling}) is in use in
a window, that forces truncation.

@defvar wrap-prefix
If this buffer-local variable is non-@code{nil}, it defines a
@dfn{wrap prefix} which Emacs displays at the start of every
continuation line.  (If lines are truncated, @code{wrap-prefix} is
never used.)  Its value may be a string or an image (@pxref{Other
Display Specs}), or a stretch of whitespace such as specified by the
@code{:width} or @code{:align-to} display properties (@pxref{Specified
Space}).  The value is interpreted in the same way as a @code{display}
text property.  @xref{Display Property}.

A wrap prefix may also be specified for regions of text, using the
@code{wrap-prefix} text or overlay property.  This takes precedence
over the @code{wrap-prefix} variable.  @xref{Special Properties}.
@end defvar

@defvar line-prefix
If this buffer-local variable is non-@code{nil}, it defines a
@dfn{line prefix} which Emacs displays at the start of every
non-continuation line.  Its value may be a string or an image
(@pxref{Other Display Specs}), or a stretch of whitespace such as
specified by the @code{:width} or @code{:align-to} display properties
(@pxref{Specified Space}).  The value is interpreted in the same way
as a @code{display} text property.  @xref{Display Property}.

A line prefix may also be specified for regions of text using the
@code{line-prefix} text or overlay property.  This takes precedence
over the @code{line-prefix} variable.  @xref{Special Properties}.
@end defvar

@ignore
  If your buffer contains only very short lines, you might find it
advisable to set @code{cache-long-scans} to @code{nil}.

@defvar cache-long-scans
If this variable is non-@code{nil} (the default), various indentation
and motion functions, and Emacs redisplay, cache the results of
scanning the buffer, and consult the cache to avoid rescanning regions
of the buffer unless they are modified.

Turning off the cache speeds up processing of short lines somewhat.

This variable is automatically buffer-local in every buffer.
@end defvar
@end ignore

@node The Echo Area
@section The Echo Area
@cindex error display
@cindex echo area

@c FIXME: Why not use @xref{Minibuffers} directly?  --xfq
  The @dfn{echo area} is used for displaying error messages
(@pxref{Errors}), for messages made with the @code{message} primitive,
and for echoing keystrokes.  It is not the same as the minibuffer,
despite the fact that the minibuffer appears (when active) in the same
place on the screen as the echo area.  @xref{Minibuffer,, The
Minibuffer, emacs, The GNU Emacs Manual}.

  Apart from the functions documented in this section, you can print
Lisp objects to the echo area by specifying @code{t} as the output
stream.  @xref{Output Streams}.

@menu
* Displaying Messages:: Explicitly displaying text in the echo area.
* Progress::            Informing user about progress of a long operation.
* Logging Messages::    Echo area messages are logged for the user.
* Echo Area Customization:: Controlling the echo area.
@end menu

@node Displaying Messages
@subsection Displaying Messages in the Echo Area
@cindex display message in echo area

  This section describes the standard functions for displaying
messages in the echo area.

@defun message format-string &rest arguments
This function displays a message in the echo area.
@var{format-string} is a format string, and @var{arguments} are the
objects for its format specifications, like in the @code{format}
function (@pxref{Formatting Strings}).  The resulting formatted string
is displayed in the echo area; if it contains @code{face} text
properties, it is displayed with the specified faces (@pxref{Faces}).
The string is also added to the @file{*Messages*} buffer, but without
text properties (@pxref{Logging Messages}).

In batch mode, the message is printed to the standard error stream,
followed by a newline.

If @var{format-string} is @code{nil} or the empty string,
@code{message} clears the echo area; if the echo area has been
expanded automatically, this brings it back to its normal size.  If
the minibuffer is active, this brings the minibuffer contents back
onto the screen immediately.

@example
@group
(message "Minibuffer depth is %d."
         (minibuffer-depth))
 @print{} Minibuffer depth is 0.
@result{} "Minibuffer depth is 0."
@end group

@group
---------- Echo Area ----------
Minibuffer depth is 0.
---------- Echo Area ----------
@end group
@end example

To automatically display a message in the echo area or in a pop-buffer,
depending on its size, use @code{display-message-or-buffer} (see below).
@end defun

@defmac with-temp-message message &rest body
This construct displays a message in the echo area temporarily, during
the execution of @var{body}.  It displays @var{message}, executes
@var{body}, then returns the value of the last body form while restoring
the previous echo area contents.
@end defmac

@defun message-or-box format-string &rest arguments
This function displays a message like @code{message}, but may display it
in a dialog box instead of the echo area.  If this function is called in
a command that was invoked using the mouse---more precisely, if
@code{last-nonmenu-event} (@pxref{Command Loop Info}) is either
@code{nil} or a list---then it uses a dialog box or pop-up menu to
display the message.  Otherwise, it uses the echo area.  (This is the
same criterion that @code{y-or-n-p} uses to make a similar decision; see
@ref{Yes-or-No Queries}.)

You can force use of the mouse or of the echo area by binding
@code{last-nonmenu-event} to a suitable value around the call.
@end defun

@defun message-box format-string &rest arguments
@anchor{message-box}
This function displays a message like @code{message}, but uses a dialog
box (or a pop-up menu) whenever that is possible.  If it is impossible
to use a dialog box or pop-up menu, because the terminal does not
support them, then @code{message-box} uses the echo area, like
@code{message}.
@end defun

@defun display-message-or-buffer message &optional buffer-name not-this-window frame
This function displays the message @var{message}, which may be either a
string or a buffer.  If it is shorter than the maximum height of the
echo area, as defined by @code{max-mini-window-height}, it is displayed
in the echo area, using @code{message}.  Otherwise,
@code{display-buffer} is used to show it in a pop-up buffer.

Returns either the string shown in the echo area, or when a pop-up
buffer is used, the window used to display it.

If @var{message} is a string, then the optional argument
@var{buffer-name} is the name of the buffer used to display it when a
pop-up buffer is used, defaulting to @file{*Message*}.  In the case
where @var{message} is a string and displayed in the echo area, it is
not specified whether the contents are inserted into the buffer anyway.

The optional arguments @var{not-this-window} and @var{frame} are as for
@code{display-buffer}, and only used if a buffer is displayed.
@end defun

@defun current-message
This function returns the message currently being displayed in the
echo area, or @code{nil} if there is none.
@end defun

@node Progress
@subsection Reporting Operation Progress
@cindex progress reporting

  When an operation can take a while to finish, you should inform the
user about the progress it makes.  This way the user can estimate
remaining time and clearly see that Emacs is busy working, not hung.
A convenient way to do this is to use a @dfn{progress reporter}.

  Here is a working example that does nothing useful:

@smallexample
(let ((progress-reporter
       (make-progress-reporter "Collecting mana for Emacs..."
                               0  500)))
  (dotimes (k 500)
    (sit-for 0.01)
    (progress-reporter-update progress-reporter k))
  (progress-reporter-done progress-reporter))
@end smallexample

@defun make-progress-reporter message &optional min-value max-value current-value min-change min-time
This function creates and returns a progress reporter object, which
you will use as an argument for the other functions listed below.  The
idea is to precompute as much data as possible to make progress
reporting very fast.

When this progress reporter is subsequently used, it will display
@var{message} in the echo area, followed by progress percentage.
@var{message} is treated as a simple string.  If you need it to depend
on a filename, for instance, use @code{format} before calling this
function.

The arguments @var{min-value} and @var{max-value} should be numbers
standing for the starting and final states of the operation.  For
instance, an operation that ``scans'' a buffer should set these to the
results of @code{point-min} and @code{point-max} correspondingly.
@var{max-value} should be greater than @var{min-value}.

Alternatively, you can set @var{min-value} and @var{max-value} to
@code{nil}.  In that case, the progress reporter does not report
process percentages; it instead displays a ``spinner'' that rotates a
notch each time you update the progress reporter.

If @var{min-value} and @var{max-value} are numbers, you can give the
argument @var{current-value} a numerical value specifying the initial
progress; if omitted, this defaults to @var{min-value}.

The remaining arguments control the rate of echo area updates.  The
progress reporter will wait for at least @var{min-change} more
percents of the operation to be completed before printing next
message; the default is one percent.  @var{min-time} specifies the
minimum time in seconds to pass between successive prints; the default
is 0.2 seconds.  (On some operating systems, the progress reporter may
handle fractions of seconds with varying precision).

This function calls @code{progress-reporter-update}, so the first
message is printed immediately.
@end defun

@defun progress-reporter-update reporter &optional value
This function does the main work of reporting progress of your
operation.  It displays the message of @var{reporter}, followed by
progress percentage determined by @var{value}.  If percentage is zero,
or close enough according to the @var{min-change} and @var{min-time}
arguments, then it is omitted from the output.

@var{reporter} must be the result of a call to
@code{make-progress-reporter}.  @var{value} specifies the current
state of your operation and must be between @var{min-value} and
@var{max-value} (inclusive) as passed to
@code{make-progress-reporter}.  For instance, if you scan a buffer,
then @var{value} should be the result of a call to @code{point}.

This function respects @var{min-change} and @var{min-time} as passed
to @code{make-progress-reporter} and so does not output new messages
on every invocation.  It is thus very fast and normally you should not
try to reduce the number of calls to it: resulting overhead will most
likely negate your effort.
@end defun

@defun progress-reporter-force-update reporter &optional value new-message
This function is similar to @code{progress-reporter-update} except
that it prints a message in the echo area unconditionally.

The first two arguments have the same meaning as for
@code{progress-reporter-update}.  Optional @var{new-message} allows
you to change the message of the @var{reporter}.  Since this function
always updates the echo area, such a change will be immediately
presented to the user.
@end defun

@defun progress-reporter-done reporter
This function should be called when the operation is finished.  It
prints the message of @var{reporter} followed by word ``done'' in the
echo area.

You should always call this function and not hope for
@code{progress-reporter-update} to print ``100%''.  Firstly, it may
never print it, there are many good reasons for this not to happen.
Secondly, ``done'' is more explicit.
@end defun

@defmac dotimes-with-progress-reporter (var count [result]) message body@dots{}
This is a convenience macro that works the same way as @code{dotimes}
does, but also reports loop progress using the functions described
above.  It allows you to save some typing.

You can rewrite the example in the beginning of this node using
this macro this way:

@example
(dotimes-with-progress-reporter
    (k 500)
    "Collecting some mana for Emacs..."
  (sit-for 0.01))
@end example
@end defmac

@node Logging Messages
@subsection Logging Messages in @file{*Messages*}
@cindex logging echo-area messages

  Almost all the messages displayed in the echo area are also recorded
in the @file{*Messages*} buffer so that the user can refer back to
them.  This includes all the messages that are output with
@code{message}.  By default, this buffer is read-only and uses the major
mode @code{messages-buffer-mode}.  Nothing prevents the user from
killing the @file{*Messages*} buffer, but the next display of a message
recreates it.  Any Lisp code that needs to access the
@file{*Messages*} buffer directly and wants to ensure that it exists
should use the function @code{messages-buffer}.

@defun messages-buffer
This function returns the @file{*Messages*} buffer.  If it does not
exist, it creates it, and switches it to @code{messages-buffer-mode}.
@end defun

@defopt message-log-max
This variable specifies how many lines to keep in the @file{*Messages*}
buffer.  The value @code{t} means there is no limit on how many lines to
keep.  The value @code{nil} disables message logging entirely.  Here's
how to display a message and prevent it from being logged:

@example
(let (message-log-max)
  (message @dots{}))
@end example
@end defopt

  To make @file{*Messages*} more convenient for the user, the logging
facility combines successive identical messages.  It also combines
successive related messages for the sake of two cases: question
followed by answer, and a series of progress messages.

  A ``question followed by an answer'' means two messages like the
ones produced by @code{y-or-n-p}: the first is @samp{@var{question}},
and the second is @samp{@var{question}...@var{answer}}.  The first
message conveys no additional information beyond what's in the second,
so logging the second message discards the first from the log.

  A ``series of progress messages'' means successive messages like
those produced by @code{make-progress-reporter}.  They have the form
@samp{@var{base}...@var{how-far}}, where @var{base} is the same each
time, while @var{how-far} varies.  Logging each message in the series
discards the previous one, provided they are consecutive.

  The functions @code{make-progress-reporter} and @code{y-or-n-p}
don't have to do anything special to activate the message log
combination feature.  It operates whenever two consecutive messages
are logged that share a common prefix ending in @samp{...}.

@node Echo Area Customization
@subsection Echo Area Customization

  These variables control details of how the echo area works.

@defvar cursor-in-echo-area
This variable controls where the cursor appears when a message is
displayed in the echo area.  If it is non-@code{nil}, then the cursor
appears at the end of the message.  Otherwise, the cursor appears at
point---not in the echo area at all.

The value is normally @code{nil}; Lisp programs bind it to @code{t}
for brief periods of time.
@end defvar

@defvar echo-area-clear-hook
This normal hook is run whenever the echo area is cleared---either by
@code{(message nil)} or for any other reason.
@end defvar

@defopt echo-keystrokes
This variable determines how much time should elapse before command
characters echo.  Its value must be an integer or floating point number,
which specifies the
number of seconds to wait before echoing.  If the user types a prefix
key (such as @kbd{C-x}) and then delays this many seconds before
continuing, the prefix key is echoed in the echo area.  (Once echoing
begins in a key sequence, all subsequent characters in the same key
sequence are echoed immediately.)

If the value is zero, then command input is not echoed.
@end defopt

@defvar message-truncate-lines
Normally, displaying a long message resizes the echo area to display
the entire message.  But if the variable @code{message-truncate-lines}
is non-@code{nil}, the echo area does not resize, and the message is
truncated to fit it.
@end defvar

  The variable @code{max-mini-window-height}, which specifies the
maximum height for resizing minibuffer windows, also applies to the
echo area (which is really a special use of the minibuffer window;
@pxref{Minibuffer Misc}).

@node Warnings
@section Reporting Warnings
@cindex warnings

  @dfn{Warnings} are a facility for a program to inform the user of a
possible problem, but continue running.

@menu
* Warning Basics::      Warnings concepts and functions to report them.
* Warning Variables::   Variables programs bind to customize their warnings.
* Warning Options::     Variables users set to control display of warnings.
* Delayed Warnings::    Deferring a warning until the end of a command.
@end menu

@node Warning Basics
@subsection Warning Basics
@cindex severity level

  Every warning has a textual message, which explains the problem for
the user, and a @dfn{severity level} which is a symbol.  Here are the
possible severity levels, in order of decreasing severity, and their
meanings:

@table @code
@item :emergency
A problem that will seriously impair Emacs operation soon
if you do not attend to it promptly.
@item :error
A report of data or circumstances that are inherently wrong.
@item :warning
A report of data or circumstances that are not inherently wrong, but
raise suspicion of a possible problem.
@item :debug
A report of information that may be useful if you are debugging.
@end table

  When your program encounters invalid input data, it can either
signal a Lisp error by calling @code{error} or @code{signal} or report
a warning with severity @code{:error}.  Signaling a Lisp error is the
easiest thing to do, but it means the program cannot continue
processing.  If you want to take the trouble to implement a way to
continue processing despite the bad data, then reporting a warning of
severity @code{:error} is the right way to inform the user of the
problem.  For instance, the Emacs Lisp byte compiler can report an
error that way and continue compiling other functions.  (If the
program signals a Lisp error and then handles it with
@code{condition-case}, the user won't see the error message; it could
show the message to the user by reporting it as a warning.)

@c FIXME: Why use "(bytecomp)" instead of "'bytecomp" or simply
@c "bytecomp" here?  The parens are part of warning-type-format but
@c not part of the warning type. --xfq
@cindex warning type
  Each warning has a @dfn{warning type} to classify it.  The type is a
list of symbols.  The first symbol should be the custom group that you
use for the program's user options.  For example, byte compiler
warnings use the warning type @code{(bytecomp)}.  You can also
subcategorize the warnings, if you wish, by using more symbols in the
list.

@defun display-warning type message &optional level buffer-name
This function reports a warning, using @var{message} as the message
and @var{type} as the warning type.  @var{level} should be the
severity level, with @code{:warning} being the default.

@var{buffer-name}, if non-@code{nil}, specifies the name of the buffer
for logging the warning.  By default, it is @file{*Warnings*}.
@end defun

@defun lwarn type level message &rest args
This function reports a warning using the value of @code{(format
@var{message} @var{args}...)} as the message in the @file{*Warnings*}
buffer.  In other respects it is equivalent to @code{display-warning}.
@end defun

@defun warn message &rest args
This function reports a warning using the value of @code{(format
@var{message} @var{args}...)} as the message, @code{(emacs)} as the
type, and @code{:warning} as the severity level.  It exists for
compatibility only; we recommend not using it, because you should
specify a specific warning type.
@end defun

@node Warning Variables
@subsection Warning Variables

  Programs can customize how their warnings appear by binding
the variables described in this section.

@defvar warning-levels
This list defines the meaning and severity order of the warning
severity levels.  Each element defines one severity level,
and they are arranged in order of decreasing severity.

Each element has the form @code{(@var{level} @var{string}
@var{function})}, where @var{level} is the severity level it defines.
@var{string} specifies the textual description of this level.
@var{string} should use @samp{%s} to specify where to put the warning
type information, or it can omit the @samp{%s} so as not to include
that information.

The optional @var{function}, if non-@code{nil}, is a function to call
with no arguments, to get the user's attention.

Normally you should not change the value of this variable.
@end defvar

@defvar warning-prefix-function
If non-@code{nil}, the value is a function to generate prefix text for
warnings.  Programs can bind the variable to a suitable function.
@code{display-warning} calls this function with the warnings buffer
current, and the function can insert text in it.  That text becomes
the beginning of the warning message.

The function is called with two arguments, the severity level and its
entry in @code{warning-levels}.  It should return a list to use as the
entry (this value need not be an actual member of
@code{warning-levels}).  By constructing this value, the function can
change the severity of the warning, or specify different handling for
a given severity level.

If the variable's value is @code{nil} then there is no function
to call.
@end defvar

@defvar warning-series
Programs can bind this variable to @code{t} to say that the next
warning should begin a series.  When several warnings form a series,
that means to leave point on the first warning of the series, rather
than keep moving it for each warning so that it appears on the last one.
The series ends when the local binding is unbound and
@code{warning-series} becomes @code{nil} again.

The value can also be a symbol with a function definition.  That is
equivalent to @code{t}, except that the next warning will also call
the function with no arguments with the warnings buffer current.  The
function can insert text which will serve as a header for the series
of warnings.

Once a series has begun, the value is a marker which points to the
buffer position in the warnings buffer of the start of the series.

The variable's normal value is @code{nil}, which means to handle
each warning separately.
@end defvar

@defvar warning-fill-prefix
When this variable is non-@code{nil}, it specifies a fill prefix to
use for filling each warning's text.
@end defvar

@defvar warning-type-format
This variable specifies the format for displaying the warning type
in the warning message.  The result of formatting the type this way
gets included in the message under the control of the string in the
entry in @code{warning-levels}.  The default value is @code{" (%s)"}.
If you bind it to @code{""} then the warning type won't appear at
all.
@end defvar

@node Warning Options
@subsection Warning Options

  These variables are used by users to control what happens
when a Lisp program reports a warning.

@defopt warning-minimum-level
This user option specifies the minimum severity level that should be
shown immediately to the user.  The default is @code{:warning}, which
means to immediately display all warnings except @code{:debug}
warnings.
@end defopt

@defopt warning-minimum-log-level
This user option specifies the minimum severity level that should be
logged in the warnings buffer.  The default is @code{:warning}, which
means to log all warnings except @code{:debug} warnings.
@end defopt

@defopt warning-suppress-types
This list specifies which warning types should not be displayed
immediately for the user.  Each element of the list should be a list
of symbols.  If its elements match the first elements in a warning
type, then that warning is not displayed immediately.
@end defopt

@defopt warning-suppress-log-types
This list specifies which warning types should not be logged in the
warnings buffer.  Each element of the list should be a list of
symbols.  If it matches the first few elements in a warning type, then
that warning is not logged.
@end defopt

@node Delayed Warnings
@subsection Delayed Warnings

Sometimes, you may wish to avoid showing a warning while a command is
running, and only show it only after the end of the command.  You can
use the variable @code{delayed-warnings-list} for this.

@defvar delayed-warnings-list
The value of this variable is a list of warnings to be displayed after
the current command has finished.  Each element must be a list

@smallexample
(@var{type} @var{message} [@var{level} [@var{buffer-name}]])
@end smallexample

@noindent
with the same form, and the same meanings, as the argument list of
@code{display-warning} (@pxref{Warning Basics}).  Immediately after
running @code{post-command-hook} (@pxref{Command Overview}), the Emacs
command loop displays all the warnings specified by this variable,
then resets it to @code{nil}.
@end defvar

  Programs which need to further customize the delayed warnings
mechanism can change the variable @code{delayed-warnings-hook}:

@defvar delayed-warnings-hook
This is a normal hook which is run by the Emacs command loop, after
@code{post-command-hook}, in order to to process and display delayed
warnings.

Its default value is a list of two functions:

@smallexample
(collapse-delayed-warnings display-delayed-warnings)
@end smallexample

@findex collapse-delayed-warnings
@findex display-delayed-warnings
@noindent
The function @code{collapse-delayed-warnings} removes repeated entries
from @code{delayed-warnings-list}.  The function
@code{display-delayed-warnings} calls @code{display-warning} on each
of the entries in @code{delayed-warnings-list}, in turn, and then sets
@code{delayed-warnings-list} to @code{nil}.
@end defvar

@node Invisible Text
@section Invisible Text

@cindex invisible text
You can make characters @dfn{invisible}, so that they do not appear on
the screen, with the @code{invisible} property.  This can be either a
text property (@pxref{Text Properties}) or an overlay property
(@pxref{Overlays}).  Cursor motion also partly ignores these
characters; if the command loop finds that point is inside a range of
invisible text after a command, it relocates point to the other side
of the text.

In the simplest case, any non-@code{nil} @code{invisible} property makes
a character invisible.  This is the default case---if you don't alter
the default value of @code{buffer-invisibility-spec}, this is how the
@code{invisible} property works.  You should normally use @code{t}
as the value of the @code{invisible} property if you don't plan
to set @code{buffer-invisibility-spec} yourself.

More generally, you can use the variable @code{buffer-invisibility-spec}
to control which values of the @code{invisible} property make text
invisible.  This permits you to classify the text into different subsets
in advance, by giving them different @code{invisible} values, and
subsequently make various subsets visible or invisible by changing the
value of @code{buffer-invisibility-spec}.

Controlling visibility with @code{buffer-invisibility-spec} is
especially useful in a program to display the list of entries in a
database.  It permits the implementation of convenient filtering
commands to view just a part of the entries in the database.  Setting
this variable is very fast, much faster than scanning all the text in
the buffer looking for properties to change.

@defvar buffer-invisibility-spec
This variable specifies which kinds of @code{invisible} properties
actually make a character invisible.  Setting this variable makes it
buffer-local.

@table @asis
@item @code{t}
A character is invisible if its @code{invisible} property is
non-@code{nil}.  This is the default.

@item a list
Each element of the list specifies a criterion for invisibility; if a
character's @code{invisible} property fits any one of these criteria,
the character is invisible.  The list can have two kinds of elements:

@table @code
@item @var{atom}
A character is invisible if its @code{invisible} property value is
@var{atom} or if it is a list with @var{atom} as a member; comparison
is done with @code{eq}.

@item (@var{atom} . t)
A character is invisible if its @code{invisible} property value is
@var{atom} or if it is a list with @var{atom} as a member; comparison
is done with @code{eq}.  Moreover, a sequence of such characters
displays as an ellipsis.
@end table
@end table
@end defvar

  Two functions are specifically provided for adding elements to
@code{buffer-invisibility-spec} and removing elements from it.

@defun add-to-invisibility-spec element
This function adds the element @var{element} to
@code{buffer-invisibility-spec}.  If @code{buffer-invisibility-spec}
was @code{t}, it changes to a list, @code{(t)}, so that text whose
@code{invisible} property is @code{t} remains invisible.
@end defun

@defun remove-from-invisibility-spec element
This removes the element @var{element} from
@code{buffer-invisibility-spec}.  This does nothing if @var{element}
is not in the list.
@end defun

  A convention for use of @code{buffer-invisibility-spec} is that a
major mode should use the mode's own name as an element of
@code{buffer-invisibility-spec} and as the value of the
@code{invisible} property:

@example
;; @r{If you want to display an ellipsis:}
(add-to-invisibility-spec '(my-symbol . t))
;; @r{If you don't want ellipsis:}
(add-to-invisibility-spec 'my-symbol)

(overlay-put (make-overlay beginning end)
             'invisible 'my-symbol)

;; @r{When done with the invisibility:}
(remove-from-invisibility-spec '(my-symbol . t))
;; @r{Or respectively:}
(remove-from-invisibility-spec 'my-symbol)
@end example

  You can check for invisibility using the following function:

@defun invisible-p pos-or-prop
If @var{pos-or-prop} is a marker or number, this function returns a
non-@code{nil} value if the text at that position is invisible.

If @var{pos-or-prop} is any other kind of Lisp object, that is taken
to mean a possible value of the @code{invisible} text or overlay
property.  In that case, this function returns a non-@code{nil} value
if that value would cause text to become invisible, based on the
current value of @code{buffer-invisibility-spec}.
@end defun

@vindex line-move-ignore-invisible
  Ordinarily, functions that operate on text or move point do not care
whether the text is invisible.  The user-level line motion commands
ignore invisible newlines if @code{line-move-ignore-invisible} is
non-@code{nil} (the default), but only because they are explicitly
programmed to do so.

  However, if a command ends with point inside or at the boundary of
invisible text, the main editing loop relocates point to one of the
two ends of the invisible text.  Emacs chooses the direction of
relocation so that it is the same as the overall movement direction of
the command; if in doubt, it prefers a position where an inserted char
would not inherit the @code{invisible} property.  Additionally, if the
text is not replaced by an ellipsis and the command only moved within
the invisible text, then point is moved one extra character so as to
try and reflect the command's movement by a visible movement of the
cursor.

  Thus, if the command moved point back to an invisible range (with the usual
stickiness), Emacs moves point back to the beginning of that range.  If the
command moved point forward into an invisible range, Emacs moves point forward
to the first visible character that follows the invisible text and then forward
one more character.

  Incremental search can make invisible overlays visible temporarily
and/or permanently when a match includes invisible text.  To enable
this, the overlay should have a non-@code{nil}
@code{isearch-open-invisible} property.  The property value should be a
function to be called with the overlay as an argument.  This function
should make the overlay visible permanently; it is used when the match
overlaps the overlay on exit from the search.

  During the search, such overlays are made temporarily visible by
temporarily modifying their invisible and intangible properties.  If you
want this to be done differently for a certain overlay, give it an
@code{isearch-open-invisible-temporary} property which is a function.
The function is called with two arguments: the first is the overlay, and
the second is @code{nil} to make the overlay visible, or @code{t} to
make it invisible again.

@node Selective Display
@section Selective Display
@c @cindex selective display   Duplicates selective-display

  @dfn{Selective display} refers to a pair of related features for
hiding certain lines on the screen.

@cindex explicit selective display
  The first variant, explicit selective display, was designed for use in a Lisp
program: it controls which lines are hidden by altering the text.  This kind of
hiding is now obsolete; instead you can get the same effect with the
@code{invisible} property (@pxref{Invisible Text}).

  In the second variant, the choice of lines to hide is made
automatically based on indentation.  This variant is designed to be a
user-level feature.

  The way you control explicit selective display is by replacing a
newline (control-j) with a carriage return (control-m).  The text that
was formerly a line following that newline is now hidden.  Strictly
speaking, it is temporarily no longer a line at all, since only
newlines can separate lines; it is now part of the previous line.

  Selective display does not directly affect editing commands.  For
example, @kbd{C-f} (@code{forward-char}) moves point unhesitatingly
into hidden text.  However, the replacement of newline characters with
carriage return characters affects some editing commands.  For
example, @code{next-line} skips hidden lines, since it searches only
for newlines.  Modes that use selective display can also define
commands that take account of the newlines, or that control which
parts of the text are hidden.

  When you write a selectively displayed buffer into a file, all the
control-m's are output as newlines.  This means that when you next read
in the file, it looks OK, with nothing hidden.  The selective display
effect is seen only within Emacs.

@defvar selective-display
This buffer-local variable enables selective display.  This means that
lines, or portions of lines, may be made hidden.

@itemize @bullet
@item
If the value of @code{selective-display} is @code{t}, then the character
control-m marks the start of hidden text; the control-m, and the rest
of the line following it, are not displayed.  This is explicit selective
display.

@item
If the value of @code{selective-display} is a positive integer, then
lines that start with more than that many columns of indentation are not
displayed.
@end itemize

When some portion of a buffer is hidden, the vertical movement
commands operate as if that portion did not exist, allowing a single
@code{next-line} command to skip any number of hidden lines.
However, character movement commands (such as @code{forward-char}) do
not skip the hidden portion, and it is possible (if tricky) to insert
or delete text in an hidden portion.

In the examples below, we show the @emph{display appearance} of the
buffer @code{foo}, which changes with the value of
@code{selective-display}.  The @emph{contents} of the buffer do not
change.

@example
@group
(setq selective-display nil)
     @result{} nil

---------- Buffer: foo ----------
1 on this column
 2on this column
  3n this column
  3n this column
 2on this column
1 on this column
---------- Buffer: foo ----------
@end group

@group
(setq selective-display 2)
     @result{} 2

---------- Buffer: foo ----------
1 on this column
 2on this column
 2on this column
1 on this column
---------- Buffer: foo ----------
@end group
@end example
@end defvar

@defopt selective-display-ellipses
If this buffer-local variable is non-@code{nil}, then Emacs displays
@samp{@dots{}} at the end of a line that is followed by hidden text.
This example is a continuation of the previous one.

@example
@group
(setq selective-display-ellipses t)
     @result{} t

---------- Buffer: foo ----------
1 on this column
 2on this column ...
 2on this column
1 on this column
---------- Buffer: foo ----------
@end group
@end example

You can use a display table to substitute other text for the ellipsis
(@samp{@dots{}}).  @xref{Display Tables}.
@end defopt

@node Temporary Displays
@section Temporary Displays

  Temporary displays are used by Lisp programs to put output into a
buffer and then present it to the user for perusal rather than for
editing.  Many help commands use this feature.

@defmac with-output-to-temp-buffer buffer-name forms@dots{}
This function executes @var{forms} while arranging to insert any output
they print into the buffer named @var{buffer-name}, which is first
created if necessary, and put into Help mode.  Finally, the buffer is
displayed in some window, but not selected.  (See the similar
form @code{with-temp-buffer-window} below.)

If the @var{forms} do not change the major mode in the output buffer,
so that it is still Help mode at the end of their execution, then
@code{with-output-to-temp-buffer} makes this buffer read-only at the
end, and also scans it for function and variable names to make them
into clickable cross-references.  @xref{Docstring hyperlinks, , Tips
for Documentation Strings}, in particular the item on hyperlinks in
documentation strings, for more details.

The string @var{buffer-name} specifies the temporary buffer, which
need not already exist.  The argument must be a string, not a buffer.
The buffer is erased initially (with no questions asked), and it is
marked as unmodified after @code{with-output-to-temp-buffer} exits.

@code{with-output-to-temp-buffer} binds @code{standard-output} to the
temporary buffer, then it evaluates the forms in @var{forms}.  Output
using the Lisp output functions within @var{forms} goes by default to
that buffer (but screen display and messages in the echo area, although
they are ``output'' in the general sense of the word, are not affected).
@xref{Output Functions}.

Several hooks are available for customizing the behavior
of this construct; they are listed below.

The value of the last form in @var{forms} is returned.

@example
@group
---------- Buffer: foo ----------
 This is the contents of foo.
---------- Buffer: foo ----------
@end group

@group
(with-output-to-temp-buffer "foo"
    (print 20)
    (print standard-output))
@result{} #<buffer foo>

---------- Buffer: foo ----------

20

#<buffer foo>

---------- Buffer: foo ----------
@end group
@end example
@end defmac

@defopt temp-buffer-show-function
If this variable is non-@code{nil}, @code{with-output-to-temp-buffer}
calls it as a function to do the job of displaying a help buffer.  The
function gets one argument, which is the buffer it should display.

It is a good idea for this function to run @code{temp-buffer-show-hook}
just as @code{with-output-to-temp-buffer} normally would, inside of
@code{save-selected-window} and with the chosen window and buffer
selected.
@end defopt

@defvar temp-buffer-setup-hook
This normal hook is run by @code{with-output-to-temp-buffer} before
evaluating @var{body}.  When the hook runs, the temporary buffer is
current.  This hook is normally set up with a function to put the
buffer in Help mode.
@end defvar

@defvar temp-buffer-show-hook
This normal hook is run by @code{with-output-to-temp-buffer} after
displaying the temporary buffer.  When the hook runs, the temporary buffer
is current, and the window it was displayed in is selected.
@end defvar

@defmac with-temp-buffer-window buffer-or-name action quit-function forms@dots{}
This macro is similar to @code{with-output-to-temp-buffer}.
Like that construct, it executes @var{forms} while arranging to insert
any output they print into the buffer named @var{buffer-or-name}.
Finally, the buffer is displayed in some window, but not selected.
Unlike @code{with-output-to-temp-buffer}, this does not switch to Help
mode.

The argument @var{buffer-or-name} specifies the temporary buffer.
It can be either a buffer, which must already exist, or a string,
in which case a buffer of that name is created if necessary.
The buffer is marked as unmodified and read-only when
@code{with-temp-buffer-window} exits.

This macro does not call @code{temp-buffer-show-function}.  Rather, it
passes the @var{action} argument to @code{display-buffer} in order to
display the buffer.

The value of the last form in @var{forms} is returned, unless the
argument @var{quit-function} is specified.  In that case,
it is called with two arguments: the window showing the buffer
and the result of @var{forms}.  The final return value is then
whatever @var{quit-function} returns.

@vindex temp-buffer-window-setup-hook
@vindex temp-buffer-window-show-hook
This macro uses the normal hooks @code{temp-buffer-window-setup-hook}
and @code{temp-buffer-window-show-hook} in place of the analogous hooks
run by @code{with-output-to-temp-buffer}.
@end defmac

@defun momentary-string-display string position &optional char message
This function momentarily displays @var{string} in the current buffer at
@var{position}.  It has no effect on the undo list or on the buffer's
modification status.

The momentary display remains until the next input event.  If the next
input event is @var{char}, @code{momentary-string-display} ignores it
and returns.  Otherwise, that event remains buffered for subsequent use
as input.  Thus, typing @var{char} will simply remove the string from
the display, while typing (say) @kbd{C-f} will remove the string from
the display and later (presumably) move point forward.  The argument
@var{char} is a space by default.

The return value of @code{momentary-string-display} is not meaningful.

If the string @var{string} does not contain control characters, you can
do the same job in a more general way by creating (and then subsequently
deleting) an overlay with a @code{before-string} property.
@xref{Overlay Properties}.

If @var{message} is non-@code{nil}, it is displayed in the echo area
while @var{string} is displayed in the buffer.  If it is @code{nil}, a
default message says to type @var{char} to continue.

In this example, point is initially located at the beginning of the
second line:

@example
@group
---------- Buffer: foo ----------
This is the contents of foo.
@point{}Second line.
---------- Buffer: foo ----------
@end group

@group
(momentary-string-display
  "**** Important Message! ****"
  (point) ?\r
  "Type RET when done reading")
@result{} t
@end group

@group
---------- Buffer: foo ----------
This is the contents of foo.
**** Important Message! ****Second line.
---------- Buffer: foo ----------

---------- Echo Area ----------
Type RET when done reading
---------- Echo Area ----------
@end group
@end example
@end defun

@node Overlays
@section Overlays
@cindex overlays
@c FIXME: mention intervals in this section?

You can use @dfn{overlays} to alter the appearance of a buffer's text on
the screen, for the sake of presentation features.  An overlay is an
object that belongs to a particular buffer, and has a specified
beginning and end.  It also has properties that you can examine and set;
these affect the display of the text within the overlay.

@cindex scalability of overlays
The visual effect of an overlay is the same as of the corresponding
text property (@pxref{Text Properties}).  However, due to a different
implementation, overlays generally don't scale well (many operations
take a time that is proportional to the number of overlays in the
buffer).  If you need to affect the visual appearance of many portions
in the buffer, we recommend using text properties.

An overlay uses markers to record its beginning and end; thus,
editing the text of the buffer adjusts the beginning and end of each
overlay so that it stays with the text.  When you create the overlay,
you can specify whether text inserted at the beginning should be
inside the overlay or outside, and likewise for the end of the overlay.

@menu
* Managing Overlays::   Creating and moving overlays.
* Overlay Properties::  How to read and set properties.
                          What properties do to the screen display.
* Finding Overlays::    Searching for overlays.
@end menu

@node Managing Overlays
@subsection Managing Overlays

  This section describes the functions to create, delete and move
overlays, and to examine their contents.  Overlay changes are not
recorded in the buffer's undo list, since the overlays are not
part of the buffer's contents.

@defun overlayp object
This function returns @code{t} if @var{object} is an overlay.
@end defun

@defun make-overlay start end &optional buffer front-advance rear-advance
This function creates and returns an overlay that belongs to
@var{buffer} and ranges from @var{start} to @var{end}.  Both @var{start}
and @var{end} must specify buffer positions; they may be integers or
markers.  If @var{buffer} is omitted, the overlay is created in the
current buffer.

The arguments @var{front-advance} and @var{rear-advance} specify the
marker insertion type for the start of the overlay and for the end of
the overlay, respectively.  @xref{Marker Insertion Types}.  If they
are both @code{nil}, the default, then the overlay extends to include
any text inserted at the beginning, but not text inserted at the end.
If @var{front-advance} is non-@code{nil}, text inserted at the
beginning of the overlay is excluded from the overlay.  If
@var{rear-advance} is non-@code{nil}, text inserted at the end of the
overlay is included in the overlay.
@end defun

@defun overlay-start overlay
This function returns the position at which @var{overlay} starts,
as an integer.
@end defun

@defun overlay-end overlay
This function returns the position at which @var{overlay} ends,
as an integer.
@end defun

@defun overlay-buffer overlay
This function returns the buffer that @var{overlay} belongs to.  It
returns @code{nil} if @var{overlay} has been deleted.
@end defun

@defun delete-overlay overlay
This function deletes @var{overlay}.  The overlay continues to exist as
a Lisp object, and its property list is unchanged, but it ceases to be
attached to the buffer it belonged to, and ceases to have any effect on
display.

A deleted overlay is not permanently disconnected.  You can give it a
position in a buffer again by calling @code{move-overlay}.
@end defun

@defun move-overlay overlay start end &optional buffer
This function moves @var{overlay} to @var{buffer}, and places its bounds
at @var{start} and @var{end}.  Both arguments @var{start} and @var{end}
must specify buffer positions; they may be integers or markers.

If @var{buffer} is omitted, @var{overlay} stays in the same buffer it
was already associated with; if @var{overlay} was deleted, it goes into
the current buffer.

The return value is @var{overlay}.

This is the only valid way to change the endpoints of an overlay.  Do
not try modifying the markers in the overlay by hand, as that fails to
update other vital data structures and can cause some overlays to be
``lost''.
@end defun

@defun remove-overlays &optional start end name value
This function removes all the overlays between @var{start} and
@var{end} whose property @var{name} has the value @var{value}.  It can
move the endpoints of the overlays in the region, or split them.

If @var{name} is omitted or @code{nil}, it means to delete all overlays in
the specified region.  If @var{start} and/or @var{end} are omitted or
@code{nil}, that means the beginning and end of the buffer respectively.
Therefore, @code{(remove-overlays)} removes all the overlays in the
current buffer.
@end defun

@defun copy-overlay overlay
This function returns a copy of @var{overlay}.  The copy has the same
endpoints and properties as @var{overlay}.  However, the marker
insertion type for the start of the overlay and for the end of the
overlay are set to their default values (@pxref{Marker Insertion
Types}).
@end defun

  Here are some examples:

@example
;; @r{Create an overlay.}
(setq foo (make-overlay 1 10))
     @result{} #<overlay from 1 to 10 in display.texi>
(overlay-start foo)
     @result{} 1
(overlay-end foo)
     @result{} 10
(overlay-buffer foo)
     @result{} #<buffer display.texi>
;; @r{Give it a property we can check later.}
(overlay-put foo 'happy t)
     @result{} t
;; @r{Verify the property is present.}
(overlay-get foo 'happy)
     @result{} t
;; @r{Move the overlay.}
(move-overlay foo 5 20)
     @result{} #<overlay from 5 to 20 in display.texi>
(overlay-start foo)
     @result{} 5
(overlay-end foo)
     @result{} 20
;; @r{Delete the overlay.}
(delete-overlay foo)
     @result{} nil
;; @r{Verify it is deleted.}
foo
     @result{} #<overlay in no buffer>
;; @r{A deleted overlay has no position.}
(overlay-start foo)
     @result{} nil
(overlay-end foo)
     @result{} nil
(overlay-buffer foo)
     @result{} nil
;; @r{Undelete the overlay.}
(move-overlay foo 1 20)
     @result{} #<overlay from 1 to 20 in display.texi>
;; @r{Verify the results.}
(overlay-start foo)
     @result{} 1
(overlay-end foo)
     @result{} 20
(overlay-buffer foo)
     @result{} #<buffer display.texi>
;; @r{Moving and deleting the overlay does not change its properties.}
(overlay-get foo 'happy)
     @result{} t
@end example

  Emacs stores the overlays of each buffer in two lists, divided
around an arbitrary ``center position''.  One list extends backwards
through the buffer from that center position, and the other extends
forwards from that center position.  The center position can be anywhere
in the buffer.

@defun overlay-recenter pos
This function recenters the overlays of the current buffer around
position @var{pos}.  That makes overlay lookup faster for positions
near @var{pos}, but slower for positions far away from @var{pos}.
@end defun

  A loop that scans the buffer forwards, creating overlays, can run
faster if you do @code{(overlay-recenter (point-max))} first.

@node Overlay Properties
@subsection Overlay Properties

  Overlay properties are like text properties in that the properties that
alter how a character is displayed can come from either source.  But in
most respects they are different.  @xref{Text Properties}, for comparison.

  Text properties are considered a part of the text; overlays and
their properties are specifically considered not to be part of the
text.  Thus, copying text between various buffers and strings
preserves text properties, but does not try to preserve overlays.
Changing a buffer's text properties marks the buffer as modified,
while moving an overlay or changing its properties does not.  Unlike
text property changes, overlay property changes are not recorded in
the buffer's undo list.

  Since more than one overlay can specify a property value for the
same character, Emacs lets you specify a priority value of each
overlay.  You should not make assumptions about which overlay will
prevail when there is a conflict and they have the same priority.

  These functions read and set the properties of an overlay:

@defun overlay-get overlay prop
This function returns the value of property @var{prop} recorded in
@var{overlay}, if any.  If @var{overlay} does not record any value for
that property, but it does have a @code{category} property which is a
symbol, that symbol's @var{prop} property is used.  Otherwise, the value
is @code{nil}.
@end defun

@defun overlay-put overlay prop value
This function sets the value of property @var{prop} recorded in
@var{overlay} to @var{value}.  It returns @var{value}.
@end defun

@defun overlay-properties overlay
This returns a copy of the property list of @var{overlay}.
@end defun

  See also the function @code{get-char-property} which checks both
overlay properties and text properties for a given character.
@xref{Examining Properties}.

  Many overlay properties have special meanings; here is a table
of them:

@table @code
@item priority
@kindex priority @r{(overlay property)}
This property's value (which should be a non-negative integer number)
determines the priority of the overlay.  No priority, or @code{nil},
means zero.

The priority matters when two or more overlays cover the same
character and both specify the same property; the one whose
@code{priority} value is larger overrides the other.  For the
@code{face} property, the higher priority overlay's value does not
completely override the other value; instead, its face attributes
override the face attributes of the lower priority @code{face}
property.

Currently, all overlays take priority over text properties.  Please
avoid using negative priority values, as we have not yet decided just
what they should mean.

@item window
@kindex window @r{(overlay property)}
If the @code{window} property is non-@code{nil}, then the overlay
applies only on that window.

@item category
@kindex category @r{(overlay property)}
If an overlay has a @code{category} property, we call it the
@dfn{category} of the overlay.  It should be a symbol.  The properties
of the symbol serve as defaults for the properties of the overlay.

@item face
@kindex face @r{(overlay property)}
This property controls the appearance of the text (@pxref{Faces}).
The value of the property can be the following:

@itemize @bullet
@item
A face name (a symbol or string).

@item
An anonymous face: a property list of the form @code{(@var{keyword}
@var{value} @dots{})}, where each @var{keyword} is a face attribute
name and @var{value} is a value for that attribute.

@item
A list of faces.  Each list element should be either a face name or an
anonymous face.  This specifies a face which is an aggregate of the
attributes of each of the listed faces.  Faces occurring earlier in
the list have higher priority.

@item
A cons cell of the form @code{(foreground-color . @var{color-name})}
or @code{(background-color . @var{color-name})}.  This specifies the
foreground or background color, similar to @code{(:foreground
@var{color-name})} or @code{(:background @var{color-name})}.  This
form is supported for backward compatibility only, and should be
avoided.
@end itemize

@item mouse-face
@kindex mouse-face @r{(overlay property)}
This property is used instead of @code{face} when the mouse is within
the range of the overlay.  However, Emacs ignores all face attributes
from this property that alter the text size (e.g., @code{:height},
@code{:weight}, and @code{:slant}).  Those attributes are always the
same as in the unhighlighted text.

@item display
@kindex display @r{(overlay property)}
This property activates various features that change the
way text is displayed.  For example, it can make text appear taller
or shorter, higher or lower, wider or narrower, or replaced with an image.
@xref{Display Property}.

@item help-echo
@kindex help-echo @r{(overlay property)}
If an overlay has a @code{help-echo} property, then when you move the
mouse onto the text in the overlay, Emacs displays a help string in the
echo area, or in the tooltip window.  For details see @ref{Text
help-echo}.

@item field
@kindex field @r{(overlay property)}
@c Copied from Special Properties.
Consecutive characters with the same @code{field} property constitute a
@emph{field}.  Some motion functions including @code{forward-word} and
@code{beginning-of-line} stop moving at a field boundary.
@xref{Fields}.

@item modification-hooks
@kindex modification-hooks @r{(overlay property)}
This property's value is a list of functions to be called if any
character within the overlay is changed or if text is inserted strictly
within the overlay.

The hook functions are called both before and after each change.
If the functions save the information they receive, and compare notes
between calls, they can determine exactly what change has been made
in the buffer text.

When called before a change, each function receives four arguments: the
overlay, @code{nil}, and the beginning and end of the text range to be
modified.

When called after a change, each function receives five arguments: the
overlay, @code{t}, the beginning and end of the text range just
modified, and the length of the pre-change text replaced by that range.
(For an insertion, the pre-change length is zero; for a deletion, that
length is the number of characters deleted, and the post-change
beginning and end are equal.)

If these functions modify the buffer, they should bind
@code{inhibit-modification-hooks} to @code{t} around doing so, to
avoid confusing the internal mechanism that calls these hooks.

Text properties also support the @code{modification-hooks} property,
but the details are somewhat different (@pxref{Special Properties}).

@item insert-in-front-hooks
@kindex insert-in-front-hooks @r{(overlay property)}
This property's value is a list of functions to be called before and
after inserting text right at the beginning of the overlay.  The calling
conventions are the same as for the @code{modification-hooks} functions.

@item insert-behind-hooks
@kindex insert-behind-hooks @r{(overlay property)}
This property's value is a list of functions to be called before and
after inserting text right at the end of the overlay.  The calling
conventions are the same as for the @code{modification-hooks} functions.

@item invisible
@kindex invisible @r{(overlay property)}
The @code{invisible} property can make the text in the overlay
invisible, which means that it does not appear on the screen.
@xref{Invisible Text}, for details.

@item intangible
@kindex intangible @r{(overlay property)}
The @code{intangible} property on an overlay works just like the
@code{intangible} text property.  @xref{Special Properties}, for details.

@item isearch-open-invisible
This property tells incremental search how to make an invisible overlay
visible, permanently, if the final match overlaps it.  @xref{Invisible
Text}.

@item isearch-open-invisible-temporary
This property tells incremental search how to make an invisible overlay
visible, temporarily, during the search.  @xref{Invisible Text}.

@item before-string
@kindex before-string @r{(overlay property)}
This property's value is a string to add to the display at the beginning
of the overlay.  The string does not appear in the buffer in any
sense---only on the screen.

@item after-string
@kindex after-string @r{(overlay property)}
This property's value is a string to add to the display at the end of
the overlay.  The string does not appear in the buffer in any
sense---only on the screen.

@item line-prefix
This property specifies a display spec to prepend to each
non-continuation line at display-time.  @xref{Truncation}.

@item wrap-prefix
This property specifies a display spec to prepend to each continuation
line at display-time.  @xref{Truncation}.

@item evaporate
@kindex evaporate @r{(overlay property)}
If this property is non-@code{nil}, the overlay is deleted automatically
if it becomes empty (i.e., if its length becomes zero).  If you give
an empty overlay a non-@code{nil} @code{evaporate} property, that deletes
it immediately.

@item keymap
@cindex keymap of character (and overlays)
@kindex keymap @r{(overlay property)}
If this property is non-@code{nil}, it specifies a keymap for a portion of the
text.  This keymap is used when the character after point is within the
overlay, and takes precedence over most other keymaps.  @xref{Active Keymaps}.

@item local-map
@kindex local-map @r{(overlay property)}
The @code{local-map} property is similar to @code{keymap} but replaces the
buffer's local map rather than augmenting existing keymaps.  This also means it
has lower precedence than minor mode keymaps.
@end table

The @code{keymap} and @code{local-map} properties do not affect a
string displayed by the @code{before-string}, @code{after-string}, or
@code{display} properties.  This is only relevant for mouse clicks and
other mouse events that fall on the string, since point is never on
the string.  To bind special mouse events for the string, assign it a
@code{keymap} or @code{local-map} text property.  @xref{Special
Properties}.

@node Finding Overlays
@subsection Searching for Overlays

@defun overlays-at pos
This function returns a list of all the overlays that cover the
character at position @var{pos} in the current buffer.  The list is in
no particular order.  An overlay contains position @var{pos} if it
begins at or before @var{pos}, and ends after @var{pos}.

To illustrate usage, here is a Lisp function that returns a list of the
overlays that specify property @var{prop} for the character at point:

@smallexample
(defun find-overlays-specifying (prop)
  (let ((overlays (overlays-at (point)))
        found)
    (while overlays
      (let ((overlay (car overlays)))
        (if (overlay-get overlay prop)
            (setq found (cons overlay found))))
      (setq overlays (cdr overlays)))
    found))
@end smallexample
@end defun

@defun overlays-in beg end
This function returns a list of the overlays that overlap the region
@var{beg} through @var{end}.  ``Overlap'' means that at least one
character is contained within the overlay and also contained within the
specified region; however, empty overlays are included in the result if
they are located at @var{beg}, strictly between @var{beg} and @var{end},
or at @var{end} when @var{end} denotes the position at the end of the
buffer.
@end defun

@defun next-overlay-change pos
This function returns the buffer position of the next beginning or end
of an overlay, after @var{pos}.  If there is none, it returns
@code{(point-max)}.
@end defun

@defun previous-overlay-change pos
This function returns the buffer position of the previous beginning or
end of an overlay, before @var{pos}.  If there is none, it returns
@code{(point-min)}.
@end defun

  As an example, here's a simplified (and inefficient) version of the
primitive function @code{next-single-char-property-change}
(@pxref{Property Search}).  It searches forward from position
@var{pos} for the next position where the value of a given property
@code{prop}, as obtained from either overlays or text properties,
changes.

@smallexample
(defun next-single-char-property-change (position prop)
  (save-excursion
    (goto-char position)
    (let ((propval (get-char-property (point) prop)))
      (while (and (not (eobp))
                  (eq (get-char-property (point) prop) propval))
        (goto-char (min (next-overlay-change (point))
                        (next-single-property-change (point) prop)))))
    (point)))
@end smallexample

@node Width
@section Width

Since not all characters have the same width, these functions let you
check the width of a character.  @xref{Primitive Indent}, and
@ref{Screen Lines}, for related functions.

@defun char-width char
This function returns the width in columns of the character
@var{char}, if it were displayed in the current buffer (i.e., taking
into account the buffer's display table, if any; @pxref{Display
Tables}).  The width of a tab character is usually @code{tab-width}
(@pxref{Usual Display}).
@end defun

@defun string-width string
This function returns the width in columns of the string @var{string},
if it were displayed in the current buffer and the selected window.
@end defun

@defun truncate-string-to-width string width &optional start-column padding ellipsis
This function returns the part of @var{string} that fits within
@var{width} columns, as a new string.

If @var{string} does not reach @var{width}, then the result ends where
@var{string} ends.  If one multi-column character in @var{string}
extends across the column @var{width}, that character is not included in
the result.  Thus, the result can fall short of @var{width} but cannot
go beyond it.

The optional argument @var{start-column} specifies the starting column.
If this is non-@code{nil}, then the first @var{start-column} columns of
the string are omitted from the value.  If one multi-column character in
@var{string} extends across the column @var{start-column}, that
character is not included.

The optional argument @var{padding}, if non-@code{nil}, is a padding
character added at the beginning and end of the result string, to extend
it to exactly @var{width} columns.  The padding character is used at the
end of the result if it falls short of @var{width}.  It is also used at
the beginning of the result if one multi-column character in
@var{string} extends across the column @var{start-column}.

If @var{ellipsis} is non-@code{nil}, it should be a string which will
replace the end of @var{string} (including any padding) if it extends
beyond @var{width}, unless the display width of @var{string} is equal
to or less than the display width of @var{ellipsis}.  If
@var{ellipsis} is non-@code{nil} and not a string, it stands for
@code{"..."}.

@example
(truncate-string-to-width "\tab\t" 12 4)
     @result{} "ab"
(truncate-string-to-width "\tab\t" 12 4 ?\s)
     @result{} "    ab  "
@end example
@end defun

@node Line Height
@section Line Height
@cindex line height
@cindex height of a line

  The total height of each display line consists of the height of the
contents of the line, plus optional additional vertical line spacing
above or below the display line.

  The height of the line contents is the maximum height of any
character or image on that display line, including the final newline
if there is one.  (A display line that is continued doesn't include a
final newline.)  That is the default line height, if you do nothing to
specify a greater height.  (In the most common case, this equals the
height of the default frame font.)

  There are several ways to explicitly specify a larger line height,
either by specifying an absolute height for the display line, or by
specifying vertical space.  However, no matter what you specify, the
actual line height can never be less than the default.

@kindex line-height @r{(text property)}
  A newline can have a @code{line-height} text or overlay property
that controls the total height of the display line ending in that
newline.

  If the property value is @code{t}, the newline character has no
effect on the displayed height of the line---the visible contents
alone determine the height.  This is useful for tiling small images
(or image slices) without adding blank areas between the images.

  If the property value is a list of the form @code{(@var{height}
@var{total})}, that adds extra space @emph{below} the display line.
First Emacs uses @var{height} as a height spec to control extra space
@emph{above} the line; then it adds enough space @emph{below} the line
to bring the total line height up to @var{total}.  In this case, the
other ways to specify the line spacing are ignored.

@cindex height spec
  Any other kind of property value is a height spec, which translates
into a number---the specified line height.  There are several ways to
write a height spec; here's how each of them translates into a number:

@table @code
@item @var{integer}
If the height spec is a positive integer, the height value is that integer.
@item @var{float}
If the height spec is a float, @var{float}, the numeric height value
is @var{float} times the frame's default line height.
@item (@var{face} . @var{ratio})
If the height spec is a cons of the format shown, the numeric height
is @var{ratio} times the height of face @var{face}.  @var{ratio} can
be any type of number, or @code{nil} which means a ratio of 1.
If @var{face} is @code{t}, it refers to the current face.
@item (nil . @var{ratio})
If the height spec is a cons of the format shown, the numeric height
is @var{ratio} times the height of the contents of the line.
@end table

  Thus, any valid height spec determines the height in pixels, one way
or another.  If the line contents' height is less than that, Emacs
adds extra vertical space above the line to achieve the specified
total height.

  If you don't specify the @code{line-height} property, the line's
height consists of the contents' height plus the line spacing.
There are several ways to specify the line spacing for different
parts of Emacs text.

  On graphical terminals, you can specify the line spacing for all
lines in a frame, using the @code{line-spacing} frame parameter
(@pxref{Layout Parameters}).  However, if the default value of
@code{line-spacing} is non-@code{nil}, it overrides the
frame's @code{line-spacing} parameter.  An integer value specifies the
number of pixels put below lines.  A floating point number specifies
the spacing relative to the frame's default line height.

@vindex line-spacing
  You can specify the line spacing for all lines in a buffer via the
buffer-local @code{line-spacing} variable.  An integer value specifies
the number of pixels put below lines.  A floating point number
specifies the spacing relative to the default frame line height.  This
overrides line spacings specified for the frame.

@kindex line-spacing @r{(text property)}
  Finally, a newline can have a @code{line-spacing} text or overlay
property that overrides the default frame line spacing and the buffer
local @code{line-spacing} variable, for the display line ending in
that newline.

  One way or another, these mechanisms specify a Lisp value for the
spacing of each line.  The value is a height spec, and it translates
into a Lisp value as described above.  However, in this case the
numeric height value specifies the line spacing, rather than the line
height.

  On text terminals, the line spacing cannot be altered.

@node Faces
@section Faces
@cindex faces

  A @dfn{face} is a collection of graphical attributes for displaying
text: font, foreground color, background color, optional underlining,
etc.  Faces control how Emacs displays text in buffers, as well as
other parts of the frame such as the mode line.

@cindex anonymous face
  One way to represent a face is as a property list of attributes,
like @code{(:foreground "red" :weight bold)}.  Such a list is called
an @dfn{anonymous face}.  For example, you can assign an anonymous
face as the value of the @code{face} text property, and Emacs will
display the underlying text with the specified attributes.
@xref{Special Properties}.

@cindex face name
  More commonly, a face is referred to via a @dfn{face name}: a Lisp
symbol associated with a set of face attributes@footnote{For backward
compatibility, you can also use a string to specify a face name; that
is equivalent to a Lisp symbol with the same name.}.  Named faces are
defined using the @code{defface} macro (@pxref{Defining Faces}).
Emacs comes with several standard named faces (@pxref{Basic Faces}).

  Many parts of Emacs required named faces, and do not accept
anonymous faces.  These include the functions documented in
@ref{Attribute Functions}, and the variable @code{font-lock-keywords}
(@pxref{Search-based Fontification}).  Unless otherwise stated, we
will use the term @dfn{face} to refer only to named faces.

@defun facep object
This function returns a non-@code{nil} value if @var{object} is a
named face: a Lisp symbol or string which serves as a face name.
Otherwise, it returns @code{nil}.
@end defun

@menu
* Face Attributes::     What is in a face?
* Defining Faces::      How to define a face.
* Attribute Functions::  Functions to examine and set face attributes.
* Displaying Faces::     How Emacs combines the faces specified for a character.
* Face Remapping::      Remapping faces to alternative definitions.
* Face Functions::      How to define and examine faces.
* Auto Faces::          Hook for automatic face assignment.
* Basic Faces::         Faces that are defined by default.
* Font Selection::      Finding the best available font for a face.
* Font Lookup::         Looking up the names of available fonts
                          and information about them.
* Fontsets::            A fontset is a collection of fonts
                          that handle a range of character sets.
* Low-Level Font::      Lisp representation for character display fonts.
@end menu

@node Face Attributes
@subsection Face Attributes
@cindex face attributes

  @dfn{Face attributes} determine the visual appearance of a face.
The following table lists all the face attributes, their possible
values, and their effects.

  Apart from the values given below, each face attribute can have the
value @code{unspecified}.  This special value means that the face
doesn't specify that attribute directly.  An @code{unspecified}
attribute tells Emacs to refer instead to a parent face (see the
description @code{:inherit} attribute below); or, failing that, to an
underlying face (@pxref{Displaying Faces}).  The @code{default} face
must specify all attributes.

  Some of these attributes are meaningful only on certain kinds of
displays.  If your display cannot handle a certain attribute, the
attribute is ignored.

@table @code
@item :family
Font family or fontset (a string).  @xref{Fonts,,, emacs, The GNU
Emacs Manual}, for more information about font families.  The function
@code{font-family-list} (see below) returns a list of available family
names.  @xref{Fontsets}, for information about fontsets.

@item :foundry
The name of the @dfn{font foundry} for the font family specified by
the @code{:family} attribute (a string).  @xref{Fonts,,, emacs, The
GNU Emacs Manual}.

@item :width
Relative character width.  This should be one of the symbols
@code{ultra-condensed}, @code{extra-condensed}, @code{condensed},
@code{semi-condensed}, @code{normal}, @code{semi-expanded},
@code{expanded}, @code{extra-expanded}, or @code{ultra-expanded}.

@item :height
The height of the font.  In the simplest case, this is an integer in
units of 1/10 point.

The value can also be a floating point number or a function, which
specifies the height relative to an @dfn{underlying face}
(@pxref{Displaying Faces}).  If the value is a floating point number,
that specifies the amount by which to scale the height of the
underlying face.  If the value is a function, that function is called
with one argument, the height of the underlying face, and returns the
height of the new face.  If the function is passed an integer
argument, it must return an integer.

The height of the default face must be specified using an integer;
floating point and function values are not allowed.

@item :weight
Font weight---one of the symbols (from densest to faintest)
@code{ultra-bold}, @code{extra-bold}, @code{bold}, @code{semi-bold},
@code{normal}, @code{semi-light}, @code{light}, @code{extra-light}, or
@code{ultra-light}.  On text terminals which support
variable-brightness text, any weight greater than normal is displayed
as extra bright, and any weight less than normal is displayed as
half-bright.

@cindex italic text
@item :slant
Font slant---one of the symbols @code{italic}, @code{oblique},
@code{normal}, @code{reverse-italic}, or @code{reverse-oblique}.  On
text terminals that support variable-brightness text, slanted text is
displayed as half-bright.

@item :foreground
Foreground color, a string.  The value can be a system-defined color
name, or a hexadecimal color specification.  @xref{Color Names}.  On
black-and-white displays, certain shades of gray are implemented by
stipple patterns.

@item :distant-foreground
Alternative foreground color, a string.  This is like @code{:foreground}
but the color is only used as a foreground when the background color is
near to the foreground that would have been used.  This is useful for
example when marking text (i.e. the region face).  If the text has a foreground
that is visible with the region face, that foreground is used.
If the foreground is near the region face background,
@code{:distant-foreground} is used instead so the text is readable.

@item :background
Background color, a string.  The value can be a system-defined color
name, or a hexadecimal color specification.  @xref{Color Names}.

@cindex underlined text
@item :underline
Whether or not characters should be underlined, and in what
way.  The possible values of the @code{:underline} attribute are:

@table @asis
@item @code{nil}
Don't underline.

@item @code{t}
Underline with the foreground color of the face.

@item @var{color}
Underline in color @var{color}, a string specifying a color.

@item @code{(:color @var{color} :style @var{style})}
@var{color} is either a string, or the symbol @code{foreground-color},
meaning the foreground color of the face.  Omitting the attribute
@code{:color} means to use the foreground color of the face.
@var{style} should be a symbol @code{line} or @code{wave}, meaning to
use a straight or wavy line.  Omitting the attribute @code{:style}
means to use a straight line.
@end table

@cindex overlined text
@item :overline
Whether or not characters should be overlined, and in what color.
If the value is @code{t}, overlining uses the foreground color of the
face.  If the value is a string, overlining uses that color.  The
value @code{nil} means do not overline.

@cindex strike-through text
@item :strike-through
Whether or not characters should be strike-through, and in what
color.  The value is used like that of @code{:overline}.

@cindex 2D box
@cindex 3D box
@item :box
Whether or not a box should be drawn around characters, its color, the
width of the box lines, and 3D appearance.  Here are the possible
values of the @code{:box} attribute, and what they mean:

@table @asis
@item @code{nil}
Don't draw a box.

@item @code{t}
Draw a box with lines of width 1, in the foreground color.

@item @var{color}
Draw a box with lines of width 1, in color @var{color}.

@item @code{(:line-width @var{width} :color @var{color} :style @var{style})}
This way you can explicitly specify all aspects of the box.  The value
@var{width} specifies the width of the lines to draw; it defaults to
1.  A negative width @var{-n} means to draw a line of width @var{n}
that occupies the space of the underlying text, thus avoiding any
increase in the character height or width.

The value @var{color} specifies the color to draw with.  The default is
the foreground color of the face for simple boxes, and the background
color of the face for 3D boxes.

The value @var{style} specifies whether to draw a 3D box.  If it is
@code{released-button}, the box looks like a 3D button that is not being
pressed.  If it is @code{pressed-button}, the box looks like a 3D button
that is being pressed.  If it is @code{nil} or omitted, a plain 2D box
is used.
@end table

@item :inverse-video
Whether or not characters should be displayed in inverse video.  The
value should be @code{t} (yes) or @code{nil} (no).

@item :stipple
The background stipple, a bitmap.

The value can be a string; that should be the name of a file containing
external-format X bitmap data.  The file is found in the directories
listed in the variable @code{x-bitmap-file-path}.

Alternatively, the value can specify the bitmap directly, with a list
of the form @code{(@var{width} @var{height} @var{data})}.  Here,
@var{width} and @var{height} specify the size in pixels, and
@var{data} is a string containing the raw bits of the bitmap, row by
row.  Each row occupies @math{(@var{width} + 7) / 8} consecutive bytes
in the string (which should be a unibyte string for best results).
This means that each row always occupies at least one whole byte.

If the value is @code{nil}, that means use no stipple pattern.

Normally you do not need to set the stipple attribute, because it is
used automatically to handle certain shades of gray.

@item :font
The font used to display the face.  Its value should be a font object.
@xref{Low-Level Font}, for information about font objects, font specs,
and font entities.

When specifying this attribute using @code{set-face-attribute}
(@pxref{Attribute Functions}), you may also supply a font spec, a font
entity, or a string.  Emacs converts such values to an appropriate
font object, and stores that font object as the actual attribute
value.  If you specify a string, the contents of the string should be
a font name (@pxref{Fonts,,, emacs, The GNU Emacs Manual}); if the
font name is an XLFD containing wildcards, Emacs chooses the first
font matching those wildcards.  Specifying this attribute also changes
the values of the @code{:family}, @code{:foundry}, @code{:width},
@code{:height}, @code{:weight}, and @code{:slant} attributes.

@cindex inheritance, for faces
@item :inherit
The name of a face from which to inherit attributes, or a list of face
names.  Attributes from inherited faces are merged into the face like
an underlying face would be, with higher priority than underlying
faces (@pxref{Displaying Faces}).  If a list of faces is used,
attributes from faces earlier in the list override those from later
faces.
@end table

@defun font-family-list &optional frame
This function returns a list of available font family names.  The
optional argument @var{frame} specifies the frame on which the text is
to be displayed; if it is @code{nil}, the selected frame is used.
@end defun

@defopt underline-minimum-offset
This variable specifies the minimum distance between the baseline and
the underline, in pixels, when displaying underlined text.
@end defopt

@defopt x-bitmap-file-path
This variable specifies a list of directories for searching
for bitmap files, for the @code{:stipple} attribute.
@end defopt

@defun bitmap-spec-p object
This returns @code{t} if @var{object} is a valid bitmap specification,
suitable for use with @code{:stipple} (see above).  It returns
@code{nil} otherwise.
@end defun

@node Defining Faces
@subsection Defining Faces

@cindex face spec
  The usual way to define a face is through the @code{defface} macro.
This macro associates a face name (a symbol) with a default @dfn{face
spec}.  A face spec is a construct which specifies what attributes a
face should have on any given terminal; for example, a face spec might
specify one foreground color on high-color terminals, and a different
foreground color on low-color terminals.

  People are sometimes tempted to create a variable whose value is a
face name.  In the vast majority of cases, this is not necessary; the
usual procedure is to define a face with @code{defface}, and then use
its name directly.

@defmac defface face spec doc [keyword value]@dots{}
This macro declares @var{face} as a named face whose default face spec
is given by @var{spec}.  You should not quote the symbol @var{face},
and it should not end in @samp{-face} (that would be redundant).  The
argument @var{doc} is a documentation string for the face.  The
additional @var{keyword} arguments have the same meanings as in
@code{defgroup} and @code{defcustom} (@pxref{Common Keywords}).

If @var{face} already has a default face spec, this macro does
nothing.

The default face spec determines @var{face}'s appearance when no
customizations are in effect (@pxref{Customization}).  If @var{face}
has already been customized (via Custom themes or via customizations
read from the init file), its appearance is determined by the custom
face spec(s), which override the default face spec @var{spec}.
However, if the customizations are subsequently removed, the
appearance of @var{face} will again be determined by its default face
spec.

As an exception, if you evaluate a @code{defface} form with
@kbd{C-M-x} in Emacs Lisp mode (@code{eval-defun}), a special feature
of @code{eval-defun} overrides any custom face specs on the face,
causing the face to reflect exactly what the @code{defface} says.

The @var{spec} argument is a @dfn{face spec}, which states how the
face should appear on different kinds of terminals.  It should be an
alist whose elements each have the form

@example
(@var{display} . @var{plist})
@end example

@noindent
@var{display} specifies a class of terminals (see below).  @var{plist}
is a property list of face attributes and their values, specifying how
the face appears on such terminals.  For backward compatibility, you
can also write an element as @code{(@var{display} @var{plist})}.

The @var{display} part of an element of @var{spec} determines which
terminals the element matches.  If more than one element of @var{spec}
matches a given terminal, the first element that matches is the one
used for that terminal.  There are three possibilities for
@var{display}:

@table @asis
@item @code{default}
This element of @var{spec} doesn't match any terminal; instead, it
specifies defaults that apply to all terminals.  This element, if
used, must be the first element of @var{spec}.  Each of the following
elements can override any or all of these defaults.

@item @code{t}
This element of @var{spec} matches all terminals.  Therefore, any
subsequent elements of @var{spec} are never used.  Normally @code{t}
is used in the last (or only) element of @var{spec}.

@item a list
If @var{display} is a list, each element should have the form
@code{(@var{characteristic} @var{value}@dots{})}.  Here
@var{characteristic} specifies a way of classifying terminals, and the
@var{value}s are possible classifications which @var{display} should
apply to.  Here are the possible values of @var{characteristic}:

@table @code
@item type
The kind of window system the terminal uses---either @code{graphic}
(any graphics-capable display), @code{x}, @code{pc} (for the MS-DOS
console), @code{w32} (for MS Windows 9X/NT/2K/XP), or @code{tty} (a
non-graphics-capable display).  @xref{Window Systems, window-system}.

@item class
What kinds of colors the terminal supports---either @code{color},
@code{grayscale}, or @code{mono}.

@item background
The kind of background---either @code{light} or @code{dark}.

@item min-colors
An integer that represents the minimum number of colors the terminal
should support.  This matches a terminal if its
@code{display-color-cells} value is at least the specified integer.

@item supports
Whether or not the terminal can display the face attributes given in
@var{value}@dots{} (@pxref{Face Attributes}).  @xref{Display Face
Attribute Testing}, for more information on exactly how this testing
is done.
@end table

If an element of @var{display} specifies more than one @var{value} for
a given @var{characteristic}, any of those values is acceptable.  If
@var{display} has more than one element, each element should specify a
different @var{characteristic}; then @emph{each} characteristic of the
terminal must match one of the @var{value}s specified for it in
@var{display}.
@end table
@end defmac

  For example, here's the definition of the standard face
@code{highlight}:

@example
(defface highlight
  '((((class color) (min-colors 88) (background light))
     :background "darkseagreen2")
    (((class color) (min-colors 88) (background dark))
     :background "darkolivegreen")
    (((class color) (min-colors 16) (background light))
     :background "darkseagreen2")
    (((class color) (min-colors 16) (background dark))
     :background "darkolivegreen")
    (((class color) (min-colors 8))
     :background "green" :foreground "black")
    (t :inverse-video t))
  "Basic face for highlighting."
  :group 'basic-faces)
@end example

  Internally, Emacs stores each face's default spec in its
@code{face-defface-spec} symbol property (@pxref{Symbol Properties}).
The @code{saved-face} property stores any face spec saved by the user
using the customization buffer; the @code{customized-face} property
stores the face spec customized for the current session, but not
saved; and the @code{theme-face} property stores an alist associating
the active customization settings and Custom themes with the face
specs for that face.  The face's documentation string is stored in the
@code{face-documentation} property.

  Normally, a face is declared just once, using @code{defface}, and
any further changes to its appearance are applied using the Customize
framework (e.g., via the Customize user interface or via the
@code{custom-set-faces} function; @pxref{Applying Customizations}), or
by face remapping (@pxref{Face Remapping}).  In the rare event that
you need to change a face spec directly from Lisp, you can use the
@code{face-spec-set} function.

@defun face-spec-set face spec &optional spec-type
This function applies @var{spec} as a face spec for @code{face}.
@var{spec} should be a face spec, as described in the above
documentation for @code{defface}.

This function also defines @var{face} as a valid face name if it is
not already one, and (re)calculates its attributes on existing frames.

@cindex override spec @r{(for a face)}
The argument @var{spec-type} determines which spec to set.  If it is
@code{nil} or @code{face-override-spec}, this function sets the
@dfn{override spec}, which overrides over all other face specs on
@var{face}.  If it is @code{customized-face} or @code{saved-face},
this function sets the customized spec or the saved custom spec.  If
it is @code{face-defface-spec}, this function sets the default face
spec (the same one set by @code{defface}).  If it is @code{reset},
this function clears out all customization specs and override specs
from @var{face} (in this case, the value of @var{spec} is ignored).
Any other value of @var{spec-type} is reserved for internal use.
@end defun

@node Attribute Functions
@subsection Face Attribute Functions

  This section describes functions for directly accessing and
modifying the attributes of a named face.

@defun face-attribute face attribute &optional frame inherit
This function returns the value of the @var{attribute} attribute for
@var{face} on @var{frame}.

If @var{frame} is @code{nil}, that means the selected frame
(@pxref{Input Focus}).  If @var{frame} is @code{t}, this function
returns the value of the specified attribute for newly-created frames
(this is normally @code{unspecified}, unless you have specified some
value using @code{set-face-attribute}; see below).

If @var{inherit} is @code{nil}, only attributes directly defined by
@var{face} are considered, so the return value may be
@code{unspecified}, or a relative value.  If @var{inherit} is
non-@code{nil}, @var{face}'s definition of @var{attribute} is merged
with the faces specified by its @code{:inherit} attribute; however the
return value may still be @code{unspecified} or relative.  If
@var{inherit} is a face or a list of faces, then the result is further
merged with that face (or faces), until it becomes specified and
absolute.

To ensure that the return value is always specified and absolute, use
a value of @code{default} for @var{inherit}; this will resolve any
unspecified or relative values by merging with the @code{default} face
(which is always completely specified).

For example,

@example
(face-attribute 'bold :weight)
     @result{} bold
@end example
@end defun

@c FIXME: Add an index for "relative face attribute", maybe here?  --xfq
@defun face-attribute-relative-p attribute value
This function returns non-@code{nil} if @var{value}, when used as the
value of the face attribute @var{attribute}, is relative.  This means
it would modify, rather than completely override, any value that comes
from a subsequent face in the face list or that is inherited from
another face.

@code{unspecified} is a relative value for all attributes.  For
@code{:height}, floating point and function values are also relative.

For example:

@example
(face-attribute-relative-p :height 2.0)
     @result{} t
@end example
@end defun

@defun face-all-attributes face &optional frame
This function returns an alist of attributes of @var{face}.  The
elements of the result are name-value pairs of the form
@w{@code{(@var{attr-name} . @var{attr-value})}}.  Optional argument
@var{frame} specifies the frame whose definition of @var{face} to
return; if omitted or @code{nil}, the returned value describes the
default attributes of @var{face} for newly created frames.
@end defun

@defun merge-face-attribute attribute value1 value2
If @var{value1} is a relative value for the face attribute
@var{attribute}, returns it merged with the underlying value
@var{value2}; otherwise, if @var{value1} is an absolute value for the
face attribute @var{attribute}, returns @var{value1} unchanged.
@end defun

  Normally, Emacs uses the face specs of each face to automatically
calculate its attributes on each frame (@pxref{Defining Faces}).  The
function @code{set-face-attribute} can override this calculation by
directly assigning attributes to a face, either on a specific frame or
for all frames.  This function is mostly intended for internal usage.

@defun set-face-attribute face frame &rest arguments
This function sets one or more attributes of @var{face} for
@var{frame}.  The attributes specifies in this way override the face
spec(s) belonging to @var{face}.

The extra arguments @var{arguments} specify the attributes to set, and
the values for them.  They should consist of alternating attribute
names (such as @code{:family} or @code{:underline}) and values.  Thus,

@example
(set-face-attribute 'foo nil :weight 'bold :slant 'italic)
@end example

@noindent
sets the attribute @code{:weight} to @code{bold} and the attribute
@code{:slant} to @code{italic}.


If @var{frame} is @code{t}, this function sets the default attributes
for newly created frames.  If @var{frame} is @code{nil}, this function
sets the attributes for all existing frames, as well as for newly
created frames.
@end defun

  The following commands and functions mostly provide compatibility
with old versions of Emacs.  They work by calling
@code{set-face-attribute}.  Values of @code{t} and @code{nil} for
their @var{frame} argument are handled just like
@code{set-face-attribute} and @code{face-attribute}.  The commands
read their arguments using the minibuffer, if called interactively.

@deffn Command set-face-foreground face color &optional frame
@deffnx Command set-face-background face color &optional frame
These set the @code{:foreground} attribute (or @code{:background}
attribute, respectively) of @var{face} to @var{color}.
@end deffn

@deffn Command set-face-stipple face pattern &optional frame
This sets the @code{:stipple} attribute of @var{face} to
@var{pattern}.
@end deffn

@deffn Command set-face-font face font &optional frame
This sets the @code{:font} attribute of @var{face} to @var{font}.
@end deffn

@defun set-face-bold face bold-p &optional frame
This sets the @code{:weight} attribute of @var{face} to @var{normal}
if @var{bold-p} is @code{nil}, and to @var{bold} otherwise.
@end defun

@defun set-face-italic face italic-p &optional frame
This sets the @code{:slant} attribute of @var{face} to @var{normal} if
@var{italic-p} is @code{nil}, and to @var{italic} otherwise.
@end defun

@defun set-face-underline face underline &optional frame
This sets the @code{:underline} attribute of @var{face} to
@var{underline}.
@end defun

@defun set-face-inverse-video face inverse-video-p &optional frame
This sets the @code{:inverse-video} attribute of @var{face} to
@var{inverse-video-p}.
@end defun

@deffn Command invert-face face &optional frame
This swaps the foreground and background colors of face @var{face}.
@end deffn

  The following functions examine the attributes of a face.  They
mostly provide compatibility with old versions of Emacs.  If you don't
specify @var{frame}, they refer to the selected frame; @code{t} refers
to the default data for new frames.  They return @code{unspecified} if
the face doesn't define any value for that attribute.  If
@var{inherit} is @code{nil}, only an attribute directly defined by the
face is returned.  If @var{inherit} is non-@code{nil}, any faces
specified by its @code{:inherit} attribute are considered as well, and
if @var{inherit} is a face or a list of faces, then they are also
considered, until a specified attribute is found.  To ensure that the
return value is always specified, use a value of @code{default} for
@var{inherit}.

@defun face-font face &optional frame
This function returns the name of the font of face @var{face}.
@end defun

@defun face-foreground face &optional frame inherit
@defunx face-background face &optional frame inherit
These functions return the foreground color (or background color,
respectively) of face @var{face}, as a string.
@end defun

@defun face-stipple face &optional frame inherit
This function returns the name of the background stipple pattern of face
@var{face}, or @code{nil} if it doesn't have one.
@end defun

@defun face-bold-p face &optional frame inherit
This function returns a non-@code{nil} value if the @code{:weight}
attribute of @var{face} is bolder than normal (i.e., one of
@code{semi-bold}, @code{bold}, @code{extra-bold}, or
@code{ultra-bold}).  Otherwise, it returns @code{nil}.
@end defun

@defun face-italic-p face &optional frame inherit
This function returns a non-@code{nil} value if the @code{:slant}
attribute of @var{face} is @code{italic} or @code{oblique}, and
@code{nil} otherwise.
@end defun

@defun face-underline-p face &optional frame inherit
This function returns non-@code{nil} if face @var{face} specifies
a non-@code{nil} @code{:underline} attribute.
@end defun

@defun face-inverse-video-p face &optional frame inherit
This function returns non-@code{nil} if face @var{face} specifies
a non-@code{nil} @code{:inverse-video} attribute.
@end defun

@node Displaying Faces
@subsection Displaying Faces

  When Emacs displays a given piece of text, the visual appearance of
the text may be determined by faces drawn from different sources.  If
these various sources together specify more than one face for a
particular character, Emacs merges the attributes of the various
faces.  Here is the order in which Emacs merges the faces, from
highest to lowest priority:

@itemize @bullet
@item
If the text consists of a special glyph, the glyph can specify a
particular face.  @xref{Glyphs}.

@item
If the text lies within an active region, Emacs highlights it using
the @code{region} face.  @xref{Standard Faces,,, emacs, The GNU Emacs
Manual}.

@item
If the text lies within an overlay with a non-@code{nil} @code{face}
property, Emacs applies the face(s) specified by that property.  If
the overlay has a @code{mouse-face} property and the mouse is ``near
enough'' to the overlay, Emacs applies the face or face attributes
specified by the @code{mouse-face} property instead.  @xref{Overlay
Properties}.

When multiple overlays cover one character, an overlay with higher
priority overrides those with lower priority.  @xref{Overlays}.

@item
If the text contains a @code{face} or @code{mouse-face} property,
Emacs applies the specified faces and face attributes.  @xref{Special
Properties}.  (This is how Font Lock mode faces are applied.
@xref{Font Lock Mode}.)

@item
If the text lies within the mode line of the selected window, Emacs
applies the @code{mode-line} face.  For the mode line of a
non-selected window, Emacs applies the @code{mode-line-inactive} face.
For a header line, Emacs applies the @code{header-line} face.

@item
If any given attribute has not been specified during the preceding
steps, Emacs applies the attribute of the @code{default} face.
@end itemize

  At each stage, if a face has a valid @code{:inherit} attribute,
Emacs treats any attribute with an @code{unspecified} value as having
the corresponding value drawn from the parent face(s).  @pxref{Face
Attributes}.  Note that the parent face(s) may also leave the
attribute unspecified; in that case, the attribute remains unspecified
at the next level of face merging.

@node Face Remapping
@subsection Face Remapping

  The variable @code{face-remapping-alist} is used for buffer-local or
global changes in the appearance of a face.  For instance, it is used
to implement the @code{text-scale-adjust} command (@pxref{Text
Scale,,, emacs, The GNU Emacs Manual}).

@defvar face-remapping-alist
The value of this variable is an alist whose elements have the form
@code{(@var{face} . @var{remapping})}.  This causes Emacs to display
any text having the face @var{face} with @var{remapping}, rather than
the ordinary definition of @var{face}.

@var{remapping} may be any face spec suitable for a @code{face} text
property: either a face (i.e., a face name or a property list of
attribute/value pairs), or a list of faces.  For details, see the
description of the @code{face} text property in @ref{Special
Properties}.  @var{remapping} serves as the complete specification for
the remapped face---it replaces the normal definition of @var{face},
instead of modifying it.

If @code{face-remapping-alist} is buffer-local, its local value takes
effect only within that buffer.

Note: face remapping is non-recursive.  If @var{remapping} references
the same face name @var{face}, either directly or via the
@code{:inherit} attribute of some other face in @var{remapping}, that
reference uses the normal definition of @var{face}.  For instance, if
the @code{mode-line} face is remapped using this entry in
@code{face-remapping-alist}:

@example
(mode-line italic mode-line)
@end example

@noindent
then the new definition of the @code{mode-line} face inherits from the
@code{italic} face, and the @emph{normal} (non-remapped) definition of
@code{mode-line} face.
@end defvar

@cindex relative remapping, faces
@cindex base remapping, faces
  The following functions implement a higher-level interface to
@code{face-remapping-alist}.  Most Lisp code should use these
functions instead of setting @code{face-remapping-alist} directly, to
avoid trampling on remappings applied elsewhere.  These functions are
intended for buffer-local remappings, so they all make
@code{face-remapping-alist} buffer-local as a side-effect. They manage
@code{face-remapping-alist} entries of the form

@example
  (@var{face} @var{relative-spec-1} @var{relative-spec-2} @var{...} @var{base-spec})
@end example

@noindent
where, as explained above, each of the @var{relative-spec-N} and
@var{base-spec} is either a face name, or a property list of
attribute/value pairs.  Each of the @dfn{relative remapping} entries,
@var{relative-spec-N}, is managed by the
@code{face-remap-add-relative} and @code{face-remap-remove-relative}
functions; these are intended for simple modifications like changing
the text size.  The @dfn{base remapping} entry, @var{base-spec}, has
the lowest priority and is managed by the @code{face-remap-set-base}
and @code{face-remap-reset-base} functions; it is intended for major
modes to remap faces in the buffers they control.

@defun face-remap-add-relative face &rest specs
This function adds the face spec in @var{specs} as relative
remappings for face @var{face} in the current buffer.  The remaining
arguments, @var{specs}, should form either a list of face names, or a
property list of attribute/value pairs.

The return value is a Lisp object that serves as a ``cookie''; you can
pass this object as an argument to @code{face-remap-remove-relative}
if you need to remove the remapping later.

@example
;; Remap the `escape-glyph' face into a combination
;; of the `highlight' and `italic' faces:
(face-remap-add-relative 'escape-glyph 'highlight 'italic)

;; Increase the size of the `default' face by 50%:
(face-remap-add-relative 'default :height 1.5)
@end example
@end defun

@defun face-remap-remove-relative cookie
This function removes a relative remapping previously added by
@code{face-remap-add-relative}.  @var{cookie} should be the Lisp
object returned by @code{face-remap-add-relative} when the remapping
was added.
@end defun

@defun face-remap-set-base face &rest specs
This function sets the base remapping of @var{face} in the current
buffer to @var{specs}.  If @var{specs} is empty, the default base
remapping is restored, similar to calling @code{face-remap-reset-base}
(see below); note that this is different from @var{specs} containing a
single value @code{nil}, which has the opposite result (the global
definition of @var{face} is ignored).

This overwrites the default @var{base-spec}, which inherits the global
face definition, so it is up to the caller to add such inheritance if
so desired.
@end defun

@defun face-remap-reset-base face
This function sets the base remapping of @var{face} to its default
value, which inherits from @var{face}'s global definition.
@end defun

@node Face Functions
@subsection Functions for Working with Faces

  Here are additional functions for creating and working with faces.

@defun face-list
This function returns a list of all defined face names.
@end defun

@defun face-id face
This function returns the @dfn{face number} of face @var{face}.  This
is a number that uniquely identifies a face at low levels within
Emacs.  It is seldom necessary to refer to a face by its face number.
@end defun

@defun face-documentation face
This function returns the documentation string of face @var{face}, or
@code{nil} if none was specified for it.
@end defun

@defun face-equal face1 face2 &optional frame
This returns @code{t} if the faces @var{face1} and @var{face2} have the
same attributes for display.
@end defun

@defun face-differs-from-default-p face &optional frame
This returns non-@code{nil} if the face @var{face} displays
differently from the default face.
@end defun

@cindex face alias
@cindex alias, for faces
A @dfn{face alias} provides an equivalent name for a face.  You can
define a face alias by giving the alias symbol the @code{face-alias}
property, with a value of the target face name.  The following example
makes @code{modeline} an alias for the @code{mode-line} face.

@example
(put 'modeline 'face-alias 'mode-line)
@end example

@defmac define-obsolete-face-alias obsolete-face current-face when
This macro defines @code{obsolete-face} as an alias for
@var{current-face}, and also marks it as obsolete, indicating that it
may be removed in future.  @var{when} should be a string indicating
when @code{obsolete-face} was made obsolete (usually a version number
string).
@end defmac

@node Auto Faces
@subsection Automatic Face Assignment
@cindex automatic face assignment
@cindex faces, automatic choice

  This hook is used for automatically assigning faces to text in the
buffer.  It is part of the implementation of Jit-Lock mode, used by
Font-Lock.

@defvar fontification-functions
This variable holds a list of functions that are called by Emacs
redisplay as needed, just before doing redisplay.  They are called even
when Font Lock Mode isn't enabled.  When Font Lock Mode is enabled, this
variable usually holds just one function, @code{jit-lock-function}.

The functions are called in the order listed, with one argument, a
buffer position @var{pos}.  Collectively they should attempt to assign
faces to the text in the current buffer starting at @var{pos}.

The functions should record the faces they assign by setting the
@code{face} property.  They should also add a non-@code{nil}
@code{fontified} property to all the text they have assigned faces to.
That property tells redisplay that faces have been assigned to that text
already.

It is probably a good idea for the functions to do nothing if the
character after @var{pos} already has a non-@code{nil} @code{fontified}
property, but this is not required.  If one function overrides the
assignments made by a previous one, the properties after the last
function finishes are the ones that really matter.

For efficiency, we recommend writing these functions so that they
usually assign faces to around 400 to 600 characters at each call.
@end defvar

@node Basic Faces
@subsection Basic Faces

If your Emacs Lisp program needs to assign some faces to text, it is
often a good idea to use certain existing faces or inherit from them,
rather than defining entirely new faces.  This way, if other users
have customized the basic faces to give Emacs a certain look, your
program will ``fit in'' without additional customization.

  Some of the basic faces defined in Emacs are listed below.  In
addition to these, you might want to make use of the Font Lock faces
for syntactic highlighting, if highlighting is not already handled by
Font Lock mode, or if some Font Lock faces are not in use.
@xref{Faces for Font Lock}.

@table @code
@item default
The default face, whose attributes are all specified.  All other faces
implicitly inherit from it: any unspecified attribute defaults to the
attribute on this face (@pxref{Face Attributes}).

@item bold
@itemx italic
@itemx bold-italic
@itemx underline
@itemx fixed-pitch
@itemx variable-pitch
These have the attributes indicated by their names (e.g., @code{bold}
has a bold @code{:weight} attribute), with all other attributes
unspecified (and so given by @code{default}).

@item shadow
For ``dimmed out'' text.  For example, it is used for the ignored
part of a filename in the minibuffer (@pxref{Minibuffer File,,
Minibuffers for File Names, emacs, The GNU Emacs Manual}).

@item link
@itemx link-visited
For clickable text buttons that send the user to a different
buffer or ``location''.

@item highlight
For stretches of text that should temporarily stand out.  For example,
it is commonly assigned to the @code{mouse-face} property for cursor
highlighting (@pxref{Special Properties}).

@item match
For text matching a search command.

@item error
@itemx warning
@itemx success
For text concerning errors, warnings, or successes.  For example,
these are used for messages in @file{*Compilation*} buffers.
@end table

@node Font Selection
@subsection Font Selection
@cindex font selection
@cindex selecting a font

  Before Emacs can draw a character on a graphical display, it must
select a @dfn{font} for that character@footnote{In this context, the
term @dfn{font} has nothing to do with Font Lock (@pxref{Font Lock
Mode}).}.  @xref{Fonts,,, emacs, The GNU Emacs Manual}.  Normally,
Emacs automatically chooses a font based on the faces assigned to that
character---specifically, the face attributes @code{:family},
@code{:weight}, @code{:slant}, and @code{:width} (@pxref{Face
Attributes}).  The choice of font also depends on the character to be
displayed; some fonts can only display a limited set of characters.
If no available font exactly fits the requirements, Emacs looks for
the @dfn{closest matching font}.  The variables in this section
control how Emacs makes this selection.

@defopt face-font-family-alternatives
If a given family is specified but does not exist, this variable
specifies alternative font families to try.  Each element should have
this form:

@example
(@var{family} @var{alternate-families}@dots{})
@end example

If @var{family} is specified but not available, Emacs will try the other
families given in @var{alternate-families}, one by one, until it finds a
family that does exist.
@end defopt

@defopt face-font-selection-order
If there is no font that exactly matches all desired face attributes
(@code{:width}, @code{:height}, @code{:weight}, and @code{:slant}),
this variable specifies the order in which these attributes should be
considered when selecting the closest matching font.  The value should
be a list containing those four attribute symbols, in order of
decreasing importance.  The default is @code{(:width :height :weight
:slant)}.

Font selection first finds the best available matches for the first
attribute in the list; then, among the fonts which are best in that
way, it searches for the best matches in the second attribute, and so
on.

The attributes @code{:weight} and @code{:width} have symbolic values in
a range centered around @code{normal}.  Matches that are more extreme
(farther from @code{normal}) are somewhat preferred to matches that are
less extreme (closer to @code{normal}); this is designed to ensure that
non-normal faces contrast with normal ones, whenever possible.

One example of a case where this variable makes a difference is when the
default font has no italic equivalent.  With the default ordering, the
@code{italic} face will use a non-italic font that is similar to the
default one.  But if you put @code{:slant} before @code{:height}, the
@code{italic} face will use an italic font, even if its height is not
quite right.
@end defopt

@defopt face-font-registry-alternatives
This variable lets you specify alternative font registries to try, if a
given registry is specified and doesn't exist.  Each element should have
this form:

@example
(@var{registry} @var{alternate-registries}@dots{})
@end example

If @var{registry} is specified but not available, Emacs will try the
other registries given in @var{alternate-registries}, one by one,
until it finds a registry that does exist.
@end defopt

@cindex scalable fonts
  Emacs can make use of scalable fonts, but by default it does not use
them.

@defopt scalable-fonts-allowed
This variable controls which scalable fonts to use.  A value of
@code{nil}, the default, means do not use scalable fonts.  @code{t}
means to use any scalable font that seems appropriate for the text.

Otherwise, the value must be a list of regular expressions.  Then a
scalable font is enabled for use if its name matches any regular
expression in the list.  For example,

@example
(setq scalable-fonts-allowed '("iso10646-1$"))
@end example

@noindent
allows the use of scalable fonts with registry @code{iso10646-1}.
@end defopt

@defvar face-font-rescale-alist
This variable specifies scaling for certain faces.  Its value should
be a list of elements of the form

@example
(@var{fontname-regexp} . @var{scale-factor})
@end example

If @var{fontname-regexp} matches the font name that is about to be
used, this says to choose a larger similar font according to the
factor @var{scale-factor}.  You would use this feature to normalize
the font size if certain fonts are bigger or smaller than their
nominal heights and widths would suggest.
@end defvar

@node Font Lookup
@subsection Looking Up Fonts

@defun x-list-fonts name &optional reference-face frame maximum width
This function returns a list of available font names that match
@var{name}.  @var{name} should be a string containing a font name in
either the Fontconfig, GTK, or XLFD format (@pxref{Fonts,,, emacs, The
GNU Emacs Manual}).  Within an XLFD string, wildcard characters may be
used: the @samp{*} character matches any substring, and the @samp{?}
character matches any single character.  Case is ignored when matching
font names.

If the optional arguments @var{reference-face} and @var{frame} are
specified, the returned list includes only fonts that are the same
size as @var{reference-face} (a face name) currently is on the frame
@var{frame}.

The optional argument @var{maximum} sets a limit on how many fonts to
return.  If it is non-@code{nil}, then the return value is truncated
after the first @var{maximum} matching fonts.  Specifying a small
value for @var{maximum} can make this function much faster, in cases
where many fonts match the pattern.

The optional argument @var{width} specifies a desired font width.  If
it is non-@code{nil}, the function only returns those fonts whose
characters are (on average) @var{width} times as wide as
@var{reference-face}.
@end defun

@defun x-family-fonts &optional family frame
This function returns a list describing the available fonts for family
@var{family} on @var{frame}.  If @var{family} is omitted or @code{nil},
this list applies to all families, and therefore, it contains all
available fonts.  Otherwise, @var{family} must be a string; it may
contain the wildcards @samp{?} and @samp{*}.

The list describes the display that @var{frame} is on; if @var{frame} is
omitted or @code{nil}, it applies to the selected frame's display
(@pxref{Input Focus}).

Each element in the list is a vector of the following form:

@example
[@var{family} @var{width} @var{point-size} @var{weight} @var{slant}
 @var{fixed-p} @var{full} @var{registry-and-encoding}]
@end example

The first five elements correspond to face attributes; if you
specify these attributes for a face, it will use this font.

The last three elements give additional information about the font.
@var{fixed-p} is non-@code{nil} if the font is fixed-pitch.
@var{full} is the full name of the font, and
@var{registry-and-encoding} is a string giving the registry and
encoding of the font.
@end defun

@node Fontsets
@subsection Fontsets

  A @dfn{fontset} is a list of fonts, each assigned to a range of
character codes.  An individual font cannot display the whole range of
characters that Emacs supports, but a fontset can.  Fontsets have names,
just as fonts do, and you can use a fontset name in place of a font name
when you specify the ``font'' for a frame or a face.  Here is
information about defining a fontset under Lisp program control.

@defun create-fontset-from-fontset-spec fontset-spec &optional style-variant-p noerror
This function defines a new fontset according to the specification
string @var{fontset-spec}.  The string should have this format:

@smallexample
@var{fontpattern}, @r{[}@var{charset}:@var{font}@r{]@dots{}}
@end smallexample

@noindent
Whitespace characters before and after the commas are ignored.

The first part of the string, @var{fontpattern}, should have the form of
a standard X font name, except that the last two fields should be
@samp{fontset-@var{alias}}.

The new fontset has two names, one long and one short.  The long name is
@var{fontpattern} in its entirety.  The short name is
@samp{fontset-@var{alias}}.  You can refer to the fontset by either
name.  If a fontset with the same name already exists, an error is
signaled, unless @var{noerror} is non-@code{nil}, in which case this
function does nothing.

If optional argument @var{style-variant-p} is non-@code{nil}, that says
to create bold, italic and bold-italic variants of the fontset as well.
These variant fontsets do not have a short name, only a long one, which
is made by altering @var{fontpattern} to indicate the bold and/or italic
status.

The specification string also says which fonts to use in the fontset.
See below for the details.
@end defun

  The construct @samp{@var{charset}:@var{font}} specifies which font to
use (in this fontset) for one particular character set.  Here,
@var{charset} is the name of a character set, and @var{font} is the font
to use for that character set.  You can use this construct any number of
times in the specification string.

  For the remaining character sets, those that you don't specify
explicitly, Emacs chooses a font based on @var{fontpattern}: it replaces
@samp{fontset-@var{alias}} with a value that names one character set.
For the @acronym{ASCII} character set, @samp{fontset-@var{alias}} is replaced
with @samp{ISO8859-1}.

  In addition, when several consecutive fields are wildcards, Emacs
collapses them into a single wildcard.  This is to prevent use of
auto-scaled fonts.  Fonts made by scaling larger fonts are not usable
for editing, and scaling a smaller font is not useful because it is
better to use the smaller font in its own size, which Emacs does.

  Thus if @var{fontpattern} is this,

@example
-*-fixed-medium-r-normal-*-24-*-*-*-*-*-fontset-24
@end example

@noindent
the font specification for @acronym{ASCII} characters would be this:

@example
-*-fixed-medium-r-normal-*-24-*-ISO8859-1
@end example

@noindent
and the font specification for Chinese GB2312 characters would be this:

@example
-*-fixed-medium-r-normal-*-24-*-gb2312*-*
@end example

  You may not have any Chinese font matching the above font
specification.  Most X distributions include only Chinese fonts that
have @samp{song ti} or @samp{fangsong ti} in the @var{family} field.  In
such a case, @samp{Fontset-@var{n}} can be specified as below:

@smallexample
Emacs.Fontset-0: -*-fixed-medium-r-normal-*-24-*-*-*-*-*-fontset-24,\
        chinese-gb2312:-*-*-medium-r-normal-*-24-*-gb2312*-*
@end smallexample

@noindent
Then, the font specifications for all but Chinese GB2312 characters have
@samp{fixed} in the @var{family} field, and the font specification for
Chinese GB2312 characters has a wild card @samp{*} in the @var{family}
field.

@defun set-fontset-font name character font-spec &optional frame add
This function modifies the existing fontset @var{name} to use the font
matching with @var{font-spec} for the character @var{character}.

If @var{name} is @code{nil}, this function modifies the fontset of the
selected frame or that of @var{frame} if @var{frame} is not
@code{nil}.

If @var{name} is @code{t}, this function modifies the default
fontset, whose short name is @samp{fontset-default}.

@var{character} may be a cons; @code{(@var{from} . @var{to})}, where
@var{from} and @var{to} are character codepoints.  In that case, use
@var{font-spec} for all characters in the range @var{from} and @var{to}
(inclusive).

@var{character} may be a charset.  In that case, use
@var{font-spec} for all character in the charsets.

@var{character} may be a script name.  In that case, use
@var{font-spec} for all character in the charsets.

@var{font-spec} may be a cons; @code{(@var{family} . @var{registry})},
where @var{family} is a family name of a font (possibly including a
foundry name at the head), @var{registry} is a registry name of a font
(possibly including an encoding name at the tail).

@var{font-spec} may be a font name string.

The optional argument @var{add}, if non-@code{nil}, specifies how to
add @var{font-spec} to the font specifications previously set.  If it
is @code{prepend}, @var{font-spec} is prepended.  If it is
@code{append}, @var{font-spec} is appended.  By default,
@var{font-spec} overrides the previous settings.

For instance, this changes the default fontset to use a font of which
family name is @samp{Kochi Gothic} for all characters belonging to
the charset @code{japanese-jisx0208}.

@smallexample
(set-fontset-font t 'japanese-jisx0208
                  (font-spec :family "Kochi Gothic"))
@end smallexample
@end defun

@defun char-displayable-p char
This function returns @code{t} if Emacs ought to be able to display
@var{char}.  More precisely, if the selected frame's fontset has a
font to display the character set that @var{char} belongs to.

Fontsets can specify a font on a per-character basis; when the fontset
does that, this function's value may not be accurate.
@end defun

@node Low-Level Font
@subsection Low-Level Font Representation
@cindex font property

  Normally, it is not necessary to manipulate fonts directly.  In case
you need to do so, this section explains how.

  In Emacs Lisp, fonts are represented using three different Lisp
object types: @dfn{font objects}, @dfn{font specs}, and @dfn{font
entities}.

@defun fontp object &optional type
Return @code{t} if @var{object} is a font object, font spec, or font
entity.  Otherwise, return @code{nil}.

The optional argument @var{type}, if non-@code{nil}, determines the
exact type of Lisp object to check for.  In that case, @var{type}
should be one of @code{font-object}, @code{font-spec}, or
@code{font-entity}.
@end defun

@cindex font object
  A font object is a Lisp object that represents a font that Emacs has
@dfn{opened}.  Font objects cannot be modified in Lisp, but they can
be inspected.

@defun font-at position &optional window string
Return the font object that is being used to display the character at
position @var{position} in the window @var{window}.  If @var{window}
is @code{nil}, it defaults to the selected window.  If @var{string} is
@code{nil}, @var{position} specifies a position in the current buffer;
otherwise, @var{string} should be a string, and @var{position}
specifies a position in that string.
@end defun

@cindex font spec
  A font spec is a Lisp object that contains a set of specifications
that can be used to find a font.  More than one font may match the
specifications in a font spec.

@defun font-spec &rest arguments
Return a new font spec using the specifications in @var{arguments},
which should come in @code{property}-@code{value} pairs.  The possible
specifications are as follows:

@table @code
@item :name
The font name (a string), in either XLFD, Fontconfig, or GTK format.
@xref{Fonts,,, emacs, The GNU Emacs Manual}.

@item :family
@itemx :foundry
@itemx :weight
@itemx :slant
@itemx :width
These have the same meanings as the face attributes of the same name.
@xref{Face Attributes}.

@item :size
The font size---either a non-negative integer that specifies the pixel
size, or a floating point number that specifies the point size.

@item :adstyle
Additional typographic style information for the font, such as
@samp{sans}.  The value should be a string or a symbol.

@cindex font registry
@item :registry
The charset registry and encoding of the font, such as
@samp{iso8859-1}.  The value should be a string or a symbol.

@item :script
The script that the font must support (a symbol).

@item :otf
@cindex OpenType font
The font must be an OpenType font that supports these OpenType
features, provided Emacs is compiled with support for @samp{libotf} (a
library for performing complex text layout in certain scripts).  The
value must be a list of the form

@smallexample
@code{(@var{script-tag} @var{langsys-tag} @var{gsub} @var{gpos})}
@end smallexample

where @var{script-tag} is the OpenType script tag symbol;
@var{langsys-tag} is the OpenType language system tag symbol, or
@code{nil} to use the default language system; @code{gsub} is a list
of OpenType GSUB feature tag symbols, or @code{nil} if none is
required; and @code{gpos} is a list of OpenType GPOS feature tag
symbols, or @code{nil} if none is required.  If @code{gsub} or
@code{gpos} is a list, a @code{nil} element in that list means that
the font must not match any of the remaining tag symbols.  The
@code{gpos} element may be omitted.
@end table
@end defun

@defun font-put font-spec property value
Set the font property @var{property} in the font-spec @var{font-spec}
to @var{value}.
@end defun

@cindex font entity
  A font entity is a reference to a font that need not be open.  Its
properties are intermediate between a font object and a font spec:
like a font object, and unlike a font spec, it refers to a single,
specific font.  Unlike a font object, creating a font entity does not
load the contents of that font into computer memory.  Emacs may open
multiple font objects of different sizes from a single font entity
referring to a scalable font.

@defun find-font font-spec &optional frame
This function returns a font entity that best matches the font spec
@var{font-spec} on frame @var{frame}.  If @var{frame} is @code{nil},
it defaults to the selected frame.
@end defun

@defun list-fonts font-spec &optional frame num prefer
This function returns a list of all font entities that match the font
spec @var{font-spec}.

The optional argument @var{frame}, if non-@code{nil}, specifies the
frame on which the fonts are to be displayed.  The optional argument
@var{num}, if non-@code{nil}, should be an integer that specifies the
maximum length of the returned list.  The optional argument
@var{prefer}, if non-@code{nil}, should be another font spec, which is
used to control the order of the returned list; the returned font
entities are sorted in order of decreasing ``closeness'' to that font
spec.
@end defun

  If you call @code{set-face-attribute} and pass a font spec, font
entity, or font name string as the value of the @code{:font}
attribute, Emacs opens the best ``matching'' font that is available
for display.  It then stores the corresponding font object as the
actual value of the @code{:font} attribute for that face.

  The following functions can be used to obtain information about a
font.  For these functions, the @var{font} argument can be a font
object, a font entity, or a font spec.

@defun font-get font property
This function returns the value of the font property @var{property}
for @var{font}.

If @var{font} is a font spec and the font spec does not specify
@var{property}, the return value is @code{nil}.  If @var{font} is a
font object or font entity, the value for the @var{:script} property
may be a list of scripts supported by the font.
@end defun

@defun font-face-attributes font &optional frame
This function returns a list of face attributes corresponding to
@var{font}.  The optional argument @var{frame} specifies the frame on
which the font is to be displayed.  If it is @code{nil}, the selected
frame is used.  The return value has the form

@smallexample
(:family @var{family} :height @var{height} :weight @var{weight}
   :slant @var{slant} :width @var{width})
@end smallexample

where the values of @var{family}, @var{height}, @var{weight},
@var{slant}, and @var{width} are face attribute values.  Some of these
key-attribute pairs may be omitted from the list if they are not
specified by @var{font}.
@end defun

@defun font-xlfd-name font &optional fold-wildcards
This function returns the XLFD (X Logical Font Descriptor), a string,
matching @var{font}.  @xref{Fonts,,, emacs, The GNU Emacs Manual}, for
information about XLFDs.  If the name is too long for an XLFD (which
can contain at most 255 characters), the function returns @code{nil}.

If the optional argument @var{fold-wildcards} is non-@code{nil},
consecutive wildcards in the XLFD are folded into one.
@end defun

@node Fringes
@section Fringes
@cindex fringes

  On graphical displays, Emacs draws @dfn{fringes} next to each
window: thin vertical strips down the sides which can display bitmaps
indicating truncation, continuation, horizontal scrolling, and so on.

@menu
* Fringe Size/Pos::     Specifying where to put the window fringes.
* Fringe Indicators::   Displaying indicator icons in the window fringes.
* Fringe Cursors::      Displaying cursors in the right fringe.
* Fringe Bitmaps::      Specifying bitmaps for fringe indicators.
* Customizing Bitmaps:: Specifying your own bitmaps to use in the fringes.
* Overlay Arrow::       Display of an arrow to indicate position.
@end menu

@node Fringe Size/Pos
@subsection Fringe Size and Position

  The following buffer-local variables control the position and width
of fringes in windows showing that buffer.

@defvar fringes-outside-margins
The fringes normally appear between the display margins and the window
text.  If the value is non-@code{nil}, they appear outside the display
margins.  @xref{Display Margins}.
@end defvar

@defvar left-fringe-width
This variable, if non-@code{nil}, specifies the width of the left
fringe in pixels.  A value of @code{nil} means to use the left fringe
width from the window's frame.
@end defvar

@defvar right-fringe-width
This variable, if non-@code{nil}, specifies the width of the right
fringe in pixels.  A value of @code{nil} means to use the right fringe
width from the window's frame.
@end defvar

  Any buffer which does not specify values for these variables uses
the values specified by the @code{left-fringe} and @code{right-fringe}
frame parameters (@pxref{Layout Parameters}).

  The above variables actually take effect via the function
@code{set-window-buffer} (@pxref{Buffers and Windows}), which calls
@code{set-window-fringes} as a subroutine.  If you change one of these
variables, the fringe display is not updated in existing windows
showing the buffer, unless you call @code{set-window-buffer} again in
each affected window.  You can also use @code{set-window-fringes} to
control the fringe display in individual windows.

@defun set-window-fringes window left &optional right outside-margins
This function sets the fringe widths of window @var{window}.
If @var{window} is @code{nil}, the selected window is used.

The argument @var{left} specifies the width in pixels of the left
fringe, and likewise @var{right} for the right fringe.  A value of
@code{nil} for either one stands for the default width.  If
@var{outside-margins} is non-@code{nil}, that specifies that fringes
should appear outside of the display margins.
@end defun

@defun window-fringes &optional window
This function returns information about the fringes of a window
@var{window}.  If @var{window} is omitted or @code{nil}, the selected
window is used.  The value has the form @code{(@var{left-width}
@var{right-width} @var{outside-margins})}.
@end defun


@node Fringe Indicators
@subsection Fringe Indicators
@cindex fringe indicators
@cindex indicators, fringe

  @dfn{Fringe indicators} are tiny icons displayed in the window
fringe to indicate truncated or continued lines, buffer boundaries,
etc.

@defopt indicate-empty-lines
@cindex fringes, and empty line indication
@cindex empty lines, indicating
When this is non-@code{nil}, Emacs displays a special glyph in the
fringe of each empty line at the end of the buffer, on graphical
displays.  @xref{Fringes}.  This variable is automatically
buffer-local in every buffer.
@end defopt

@defopt indicate-buffer-boundaries
@cindex buffer boundaries, indicating
This buffer-local variable controls how the buffer boundaries and
window scrolling are indicated in the window fringes.

Emacs can indicate the buffer boundaries---that is, the first and last
line in the buffer---with angle icons when they appear on the screen.
In addition, Emacs can display an up-arrow in the fringe to show
that there is text above the screen, and a down-arrow to show
there is text below the screen.

There are three kinds of basic values:

@table @asis
@item @code{nil}
Don't display any of these fringe icons.
@item @code{left}
Display the angle icons and arrows in the left fringe.
@item @code{right}
Display the angle icons and arrows in the right fringe.
@item any non-alist
Display the angle icons in the left fringe
and don't display the arrows.
@end table

Otherwise the value should be an alist that specifies which fringe
indicators to display and where.  Each element of the alist should
have the form @code{(@var{indicator} . @var{position})}.  Here,
@var{indicator} is one of @code{top}, @code{bottom}, @code{up},
@code{down}, and @code{t} (which covers all the icons not yet
specified), while @var{position} is one of @code{left}, @code{right}
and @code{nil}.

For example, @code{((top . left) (t . right))} places the top angle
bitmap in left fringe, and the bottom angle bitmap as well as both
arrow bitmaps in right fringe.  To show the angle bitmaps in the left
fringe, and no arrow bitmaps, use @code{((top .  left) (bottom . left))}.
@end defopt

@defvar fringe-indicator-alist
This buffer-local variable specifies the mapping from logical fringe
indicators to the actual bitmaps displayed in the window fringes.  The
value is an alist of elements @code{(@var{indicator}
. @var{bitmaps})}, where @var{indicator} specifies a logical indicator
type and @var{bitmaps} specifies the fringe bitmaps to use for that
indicator.

  Each @var{indicator} should be one of the following symbols:

@table @asis
@item @code{truncation}, @code{continuation}.
Used for truncation and continuation lines.

@item @code{up}, @code{down}, @code{top}, @code{bottom}, @code{top-bottom}
Used when @code{indicate-buffer-boundaries} is non-@code{nil}:
@code{up} and @code{down} indicate a buffer boundary lying above or
below the window edge; @code{top} and @code{bottom} indicate the
topmost and bottommost buffer text line; and @code{top-bottom}
indicates where there is just one line of text in the buffer.

@item @code{empty-line}
Used to indicate empty lines when @code{indicate-empty-lines} is
non-@code{nil}.

@item @code{overlay-arrow}
Used for overlay arrows (@pxref{Overlay Arrow}).
@c Is this used anywhere?
@c @item Unknown bitmap indicator:
@c @code{unknown}.
@end table

  Each @var{bitmaps} value may be a list of symbols @code{(@var{left}
@var{right} [@var{left1} @var{right1}])}.  The @var{left} and
@var{right} symbols specify the bitmaps shown in the left and/or right
fringe, for the specific indicator.  @var{left1} and @var{right1} are
specific to the @code{bottom} and @code{top-bottom} indicators, and
are used to indicate that the last text line has no final newline.
Alternatively, @var{bitmaps} may be a single symbol which is used in
both left and right fringes.

  @xref{Fringe Bitmaps}, for a list of standard bitmap symbols and how
to define your own.  In addition, @code{nil} represents the empty
bitmap (i.e., an indicator that is not shown).

  When @code{fringe-indicator-alist} has a buffer-local value, and
there is no bitmap defined for a logical indicator, or the bitmap is
@code{t}, the corresponding value from the default value of
@code{fringe-indicator-alist} is used.
@end defvar

@node Fringe Cursors
@subsection Fringe Cursors
@cindex fringe cursors
@cindex cursor, fringe

  When a line is exactly as wide as the window, Emacs displays the
cursor in the right fringe instead of using two lines.  Different
bitmaps are used to represent the cursor in the fringe depending on
the current buffer's cursor type.

@defopt overflow-newline-into-fringe
If this is non-@code{nil}, lines exactly as wide as the window (not
counting the final newline character) are not continued.  Instead,
when point is at the end of the line, the cursor appears in the right
fringe.
@end defopt

@defvar fringe-cursor-alist
This variable specifies the mapping from logical cursor type to the
actual fringe bitmaps displayed in the right fringe.  The value is an
alist where each element has the form @code{(@var{cursor-type}
. @var{bitmap})}, which means to use the fringe bitmap @var{bitmap} to
display cursors of type @var{cursor-type}.

Each @var{cursor-type} should be one of @code{box}, @code{hollow},
@code{bar}, @code{hbar}, or @code{hollow-small}.  The first four have
the same meanings as in the @code{cursor-type} frame parameter
(@pxref{Cursor Parameters}).  The @code{hollow-small} type is used
instead of @code{hollow} when the normal @code{hollow-rectangle}
bitmap is too tall to fit on a specific display line.

Each @var{bitmap} should be a symbol specifying the fringe bitmap to
be displayed for that logical cursor type.
@iftex
See the next subsection for details.
@end iftex
@ifnottex
@xref{Fringe Bitmaps}.
@end ifnottex

@c FIXME: I can't find the fringes-indicator-alist variable.  Maybe
@c it should be fringe-indicator-alist or fringe-cursor-alist?  --xfq
When @code{fringe-cursor-alist} has a buffer-local value, and there is
no bitmap defined for a cursor type, the corresponding value from the
default value of @code{fringes-indicator-alist} is used.
@end defvar

@node Fringe Bitmaps
@subsection Fringe Bitmaps
@cindex fringe bitmaps
@cindex bitmaps, fringe

  The @dfn{fringe bitmaps} are the actual bitmaps which represent the
logical fringe indicators for truncated or continued lines, buffer
boundaries, overlay arrows, etc.  Each bitmap is represented by a
symbol.
@iftex
These symbols are referred to by the variables
@code{fringe-indicator-alist} and @code{fringe-cursor-alist},
described in the previous subsections.
@end iftex
@ifnottex
These symbols are referred to by the variable
@code{fringe-indicator-alist}, which maps fringe indicators to bitmaps
(@pxref{Fringe Indicators}), and the variable
@code{fringe-cursor-alist}, which maps fringe cursors to bitmaps
(@pxref{Fringe Cursors}).
@end ifnottex

  Lisp programs can also directly display a bitmap in the left or
right fringe, by using a @code{display} property for one of the
characters appearing in the line (@pxref{Other Display Specs}).  Such
a display specification has the form

@example
(@var{fringe} @var{bitmap} [@var{face}])
@end example

@noindent
@var{fringe} is either the symbol @code{left-fringe} or
@code{right-fringe}.  @var{bitmap} is a symbol identifying the bitmap
to display.  The optional @var{face} names a face whose foreground
color is used to display the bitmap; this face is automatically merged
with the @code{fringe} face.

  Here is a list of the standard fringe bitmaps defined in Emacs, and
how they are currently used in Emacs (via
@code{fringe-indicator-alist} and @code{fringe-cursor-alist}):

@table @asis
@item @code{left-arrow}, @code{right-arrow}
Used to indicate truncated lines.

@item @code{left-curly-arrow}, @code{right-curly-arrow}
Used to indicate continued lines.

@item @code{right-triangle}, @code{left-triangle}
The former is used by overlay arrows.  The latter is unused.

@item @code{up-arrow}, @code{down-arrow}, @code{top-left-angle} @code{top-right-angle}
@itemx @code{bottom-left-angle}, @code{bottom-right-angle}
@itemx @code{top-right-angle}, @code{top-left-angle}
@itemx @code{left-bracket}, @code{right-bracket}, @code{top-right-angle}, @code{top-left-angle}
Used to indicate buffer boundaries.

@item @code{filled-rectangle}, @code{hollow-rectangle}
@itemx @code{filled-square}, @code{hollow-square}
@itemx @code{vertical-bar}, @code{horizontal-bar}
Used for different types of fringe cursors.

@item @code{empty-line}, @code{exclamation-mark}, @code{question-mark}, @code{exclamation-mark}
Not used by core Emacs features.
@end table

@noindent
The next subsection describes how to define your own fringe bitmaps.

@defun fringe-bitmaps-at-pos &optional pos window
This function returns the fringe bitmaps of the display line
containing position @var{pos} in window @var{window}.  The return
value has the form @code{(@var{left} @var{right} @var{ov})}, where @var{left}
is the symbol for the fringe bitmap in the left fringe (or @code{nil}
if no bitmap), @var{right} is similar for the right fringe, and @var{ov}
is non-@code{nil} if there is an overlay arrow in the left fringe.

The value is @code{nil} if @var{pos} is not visible in @var{window}.
If @var{window} is @code{nil}, that stands for the selected window.
If @var{pos} is @code{nil}, that stands for the value of point in
@var{window}.
@end defun

@node Customizing Bitmaps
@subsection Customizing Fringe Bitmaps
@cindex fringe bitmaps, customizing

@defun define-fringe-bitmap bitmap bits &optional height width align
This function defines the symbol @var{bitmap} as a new fringe bitmap,
or replaces an existing bitmap with that name.

The argument @var{bits} specifies the image to use.  It should be
either a string or a vector of integers, where each element (an
integer) corresponds to one row of the bitmap.  Each bit of an integer
corresponds to one pixel of the bitmap, where the low bit corresponds
to the rightmost pixel of the bitmap.

The height is normally the length of @var{bits}.  However, you
can specify a different height with non-@code{nil} @var{height}.  The width
is normally 8, but you can specify a different width with non-@code{nil}
@var{width}.  The width must be an integer between 1 and 16.

The argument @var{align} specifies the positioning of the bitmap
relative to the range of rows where it is used; the default is to
center the bitmap.  The allowed values are @code{top}, @code{center},
or @code{bottom}.

The @var{align} argument may also be a list @code{(@var{align}
@var{periodic})} where @var{align} is interpreted as described above.
If @var{periodic} is non-@code{nil}, it specifies that the rows in
@code{bits} should be repeated enough times to reach the specified
height.
@end defun

@defun destroy-fringe-bitmap bitmap
This function destroy the fringe bitmap identified by @var{bitmap}.
If @var{bitmap} identifies a standard fringe bitmap, it actually
restores the standard definition of that bitmap, instead of
eliminating it entirely.
@end defun

@defun set-fringe-bitmap-face bitmap &optional face
This sets the face for the fringe bitmap @var{bitmap} to @var{face}.
If @var{face} is @code{nil}, it selects the @code{fringe} face.  The
bitmap's face controls the color to draw it in.

@var{face} is merged with the @code{fringe} face, so normally
@var{face} should specify only the foreground color.
@end defun

@node Overlay Arrow
@subsection The Overlay Arrow
@c @cindex overlay arrow  Duplicates variable names

  The @dfn{overlay arrow} is useful for directing the user's attention
to a particular line in a buffer.  For example, in the modes used for
interface to debuggers, the overlay arrow indicates the line of code
about to be executed.  This feature has nothing to do with
@dfn{overlays} (@pxref{Overlays}).

@defvar overlay-arrow-string
This variable holds the string to display to call attention to a
particular line, or @code{nil} if the arrow feature is not in use.
On a graphical display the contents of the string are ignored; instead a
glyph is displayed in the fringe area to the left of the display area.
@end defvar

@defvar overlay-arrow-position
This variable holds a marker that indicates where to display the overlay
arrow.  It should point at the beginning of a line.  On a non-graphical
display the arrow text
appears at the beginning of that line, overlaying any text that would
otherwise appear.  Since the arrow is usually short, and the line
usually begins with indentation, normally nothing significant is
overwritten.

The overlay-arrow string is displayed in any given buffer if the value
of @code{overlay-arrow-position} in that buffer points into that
buffer.  Thus, it is possible to display multiple overlay arrow strings
by creating buffer-local bindings of @code{overlay-arrow-position}.
However, it is usually cleaner to use
@code{overlay-arrow-variable-list} to achieve this result.
@c !!! overlay-arrow-position: but the overlay string may remain in the display
@c of some other buffer until an update is required.  This should be fixed
@c now.  Is it?
@end defvar

  You can do a similar job by creating an overlay with a
@code{before-string} property.  @xref{Overlay Properties}.

  You can define multiple overlay arrows via the variable
@code{overlay-arrow-variable-list}.

@defvar overlay-arrow-variable-list
This variable's value is a list of variables, each of which specifies
the position of an overlay arrow.  The variable
@code{overlay-arrow-position} has its normal meaning because it is on
this list.
@end defvar

Each variable on this list can have properties
@code{overlay-arrow-string} and @code{overlay-arrow-bitmap} that
specify an overlay arrow string (for text terminals) or fringe bitmap
(for graphical terminals) to display at the corresponding overlay
arrow position.  If either property is not set, the default
@code{overlay-arrow-string} or @code{overlay-arrow} fringe indicator
is used.

@node Scroll Bars
@section Scroll Bars
@cindex scroll bars

Normally the frame parameter @code{vertical-scroll-bars} controls
whether the windows in the frame have vertical scroll bars, and
whether they are on the left or right.  The frame parameter
@code{scroll-bar-width} specifies how wide they are (@code{nil}
meaning the default).  @xref{Layout Parameters}.

@defun frame-current-scroll-bars &optional frame
This function reports the scroll bar type settings for frame
@var{frame}.  The value is a cons cell
@code{(@var{vertical-type} .@: @var{horizontal-type})}, where
@var{vertical-type} is either @code{left}, @code{right}, or @code{nil}
(which means no scroll bar.)  @var{horizontal-type} is meant to
specify the horizontal scroll bar type, but since they are not
implemented, it is always @code{nil}.
@end defun

@vindex vertical-scroll-bar
  You can enable or disable scroll bars for a particular buffer,
by setting the variable @code{vertical-scroll-bar}.  This variable
automatically becomes buffer-local when set.  The possible values are
@code{left}, @code{right}, @code{t}, which means to use the
frame's default, and @code{nil} for no scroll bar.

  You can also control this for individual windows.  Call the function
@code{set-window-scroll-bars} to specify what to do for a specific window:

@defun set-window-scroll-bars window width &optional vertical-type horizontal-type
This function sets the width and type of scroll bars for window
@var{window}.

@var{width} specifies the scroll bar width in pixels (@code{nil} means
use the width specified for the frame).  @var{vertical-type} specifies
whether to have a vertical scroll bar and, if so, where.  The possible
values are @code{left}, @code{right} and @code{nil}, just like the
values of the @code{vertical-scroll-bars} frame parameter.

The argument @var{horizontal-type} is meant to specify whether and
where to have horizontal scroll bars, but since they are not
implemented, it has no effect.  If @var{window} is @code{nil}, the
selected window is used.
@end defun

@defun window-scroll-bars &optional window
Report the width and type of scroll bars specified for @var{window}.
If @var{window} is omitted or @code{nil}, the selected window is used.
The value is a list of the form @code{(@var{width}
@var{cols} @var{vertical-type} @var{horizontal-type})}.  The value
@var{width} is the value that was specified for the width (which may
be @code{nil}); @var{cols} is the number of columns that the scroll
bar actually occupies.

@var{horizontal-type} is not actually meaningful.
@end defun

@defun window-scroll-bar-width &optional window
This function returns the width of @var{window}'s vertical scrollbar,
in pixels.  @var{window} must be a live window.  If @var{window} is
@code{nil} or omitted, it will be the selected window.
@end defun

If you don't specify these values for a window with
@code{set-window-scroll-bars}, the buffer-local variables
@code{scroll-bar-mode} and @code{scroll-bar-width} in the buffer being
displayed control the window's vertical scroll bars.  The function
@code{set-window-buffer} examines these variables.  If you change them
in a buffer that is already visible in a window, you can make the
window take note of the new values by calling @code{set-window-buffer}
specifying the same buffer that is already displayed.

@defopt scroll-bar-mode
This variable, always local in all buffers, controls whether and where
to put scroll bars in windows displaying the buffer.  The possible values
are @code{nil} for no scroll bar, @code{left} to put a scroll bar on
the left, and @code{right} to put a scroll bar on the right.
@end defopt

@defun window-current-scroll-bars &optional window
This function reports the scroll bar type for window @var{window}.
If @var{window} is omitted or @code{nil}, the selected window is used.
The value is a cons cell
@code{(@var{vertical-type} .@: @var{horizontal-type})}.  Unlike
@code{window-scroll-bars}, this reports the scroll bar type actually
used, once frame defaults and @code{scroll-bar-mode} are taken into
account.
@end defun

@defvar scroll-bar-width
This variable, always local in all buffers, specifies the width of the
buffer's scroll bars, measured in pixels.  A value of @code{nil} means
to use the value specified by the frame.
@end defvar

@node Window Dividers
@section Window Dividers
@cindex window dividers
@cindex right dividers
@cindex bottom dividers

Window dividers are bars drawn between a frame's windows.  A ``right''
divider is drawn between a window and its sibling(s) on the right.  Its
width is specified by the frame parameter @code{right-divider-width}.  A
``bottom'' divider is drawn between a window and its sibling(s) on the
bottom or the echo area.  Its width is specified by the frame parameter
@code{bottom-divider-width}.  In either case, specifying a width of zero
means to not draw such dividers.  @xref{Layout Parameters}.

   Technically, a right divider ``belongs'' to the window on its left,
which means that its width is part of the total width of that window.  A
bottom divider ``belongs'' to the window above it, which means that its
height is part of the total height of that window.  @xref{Window Sizes}.
When a window has both, a right and a bottom divider, the bottom divider
``prevails''.  This means that the width of the bottom divider equals
the total width of the window while the height of the right divider
equals the total height of the window minus the height of the bottom
divider.

   Dividers can be dragged with the mouse and are therefore useful for
adjusting the sizes of adjacent windows with the mouse.  They also serve
to set windows visually apart from their siblings when no scroll bars or
mode lines are present.  The following three faces allow to customize
the appearance of dividers:

@table @code
@item window-divider
When a divider is less than three pixels wide, it is drawn solidly with
the foreground of this face.  For larger dividers this face is used for
the inner part only, excluding the first and last pixel.

@item window-divider-first-pixel
This is the face used for drawing the first pixel of a divider that is
at least three pixels wide.  To obtain a solid appearance, set this to
the same value used for the @code{window-divider} face.

@item window-divider-last-pixel
This is the face used for drawing the last pixel of a divider that is at
least three pixels wide.  To obtain a solid appearance, set this to the
same value used for the @code{window-divider} face.
@end table

@node Display Property
@section The @code{display} Property
@cindex display specification
@kindex display @r{(text property)}

  The @code{display} text property (or overlay property) is used to
insert images into text, and to control other aspects of how text
displays.  The value of the @code{display} property should be a
display specification, or a list or vector containing several display
specifications.  Display specifications in the same @code{display}
property value generally apply in parallel to the text they cover.

  If several sources (overlays and/or a text property) specify values
for the @code{display} property, only one of the values takes effect,
following the rules of @code{get-char-property}.  @xref{Examining
Properties}.

  The rest of this section describes several kinds of
display specifications and what they mean.

@menu
* Replacing Specs::      Display specs that replace the text.
* Specified Space::      Displaying one space with a specified width.
* Pixel Specification::  Specifying space width or height in pixels.
* Other Display Specs::     Displaying an image; adjusting the height,
                              spacing, and other properties of text.
* Display Margins::     Displaying text or images to the side of the main text.
@end menu

@node Replacing Specs
@subsection Display Specs That Replace The Text

  Some kinds of display specifications specify something to display
instead of the text that has the property.  These are called
@dfn{replacing} display specifications.  Emacs does not allow the user
to interactively move point into the middle of buffer text that is
replaced in this way.

  If a list of display specifications includes more than one replacing
display specification, the first overrides the rest.  Replacing
display specifications make most other display specifications
irrelevant, since those don't apply to the replacement.

  For replacing display specifications, ``the text that has the
property'' means all the consecutive characters that have the same
Lisp object as their @code{display} property; these characters are
replaced as a single unit.  If two characters have different Lisp
objects as their @code{display} properties (i.e., objects which are
not @code{eq}), they are handled separately.

  Here is an example which illustrates this point.  A string serves as
a replacing display specification, which replaces the text that has
the property with the specified string (@pxref{Other Display Specs}).
Consider the following function:

@smallexample
(defun foo ()
  (dotimes (i 5)
    (let ((string (concat "A"))
          (start (+ i i (point-min))))
      (put-text-property start (1+ start) 'display string)
      (put-text-property start (+ 2 start) 'display string))))
@end smallexample

@noindent
This function gives each of the first ten characters in the buffer a
@code{display} property which is a string @code{"A"}, but they don't
all get the same string object.  The first two characters get the same
string object, so they are replaced with one @samp{A}; the fact that
the display property was assigned in two separate calls to
@code{put-text-property} is irrelevant.  Similarly, the next two
characters get a second string (@code{concat} creates a new string
object), so they are replaced with one @samp{A}; and so on.  Thus, the
ten characters appear as five A's.

@node Specified Space
@subsection Specified Spaces
@cindex spaces, specified height or width
@cindex variable-width spaces

  To display a space of specified width and/or height, use a display
specification of the form @code{(space . @var{props})}, where
@var{props} is a property list (a list of alternating properties and
values).  You can put this property on one or more consecutive
characters; a space of the specified height and width is displayed in
place of @emph{all} of those characters.  These are the properties you
can use in @var{props} to specify the weight of the space:

@table @code
@item :width @var{width}
If @var{width} is an integer or floating point number, it specifies
that the space width should be @var{width} times the normal character
width.  @var{width} can also be a @dfn{pixel width} specification
(@pxref{Pixel Specification}).

@item :relative-width @var{factor}
Specifies that the width of the stretch should be computed from the
first character in the group of consecutive characters that have the
same @code{display} property.  The space width is the width of that
character, multiplied by @var{factor}.

@item :align-to @var{hpos}
Specifies that the space should be wide enough to reach @var{hpos}.
If @var{hpos} is a number, it is measured in units of the normal
character width.  @var{hpos} can also be a @dfn{pixel width}
specification (@pxref{Pixel Specification}).
@end table

  You should use one and only one of the above properties.  You can
also specify the height of the space, with these properties:

@table @code
@item :height @var{height}
Specifies the height of the space.
If @var{height} is an integer or floating point number, it specifies
that the space height should be @var{height} times the normal character
height.  The @var{height} may also be a @dfn{pixel height} specification
(@pxref{Pixel Specification}).

@item :relative-height @var{factor}
Specifies the height of the space, multiplying the ordinary height
of the text having this display specification by @var{factor}.

@item :ascent @var{ascent}
If the value of @var{ascent} is a non-negative number no greater than
100, it specifies that @var{ascent} percent of the height of the space
should be considered as the ascent of the space---that is, the part
above the baseline.  The ascent may also be specified in pixel units
with a @dfn{pixel ascent} specification (@pxref{Pixel Specification}).

@end table

  Don't use both @code{:height} and @code{:relative-height} together.

  The @code{:width} and @code{:align-to} properties are supported on
non-graphic terminals, but the other space properties in this section
are not.

  Note that space properties are treated as paragraph separators for
the purposes of reordering bidirectional text for display.
@xref{Bidirectional Display}, for the details.

@node Pixel Specification
@subsection Pixel Specification for Spaces
@cindex spaces, pixel specification

  The value of the @code{:width}, @code{:align-to}, @code{:height},
and @code{:ascent} properties can be a special kind of expression that
is evaluated during redisplay.  The result of the evaluation is used
as an absolute number of pixels.

  The following expressions are supported:

@smallexample
@group
  @var{expr} ::= @var{num} | (@var{num}) | @var{unit} | @var{elem} | @var{pos} | @var{image} | @var{form}
  @var{num}  ::= @var{integer} | @var{float} | @var{symbol}
  @var{unit} ::= in | mm | cm | width | height
@end group
@group
  @var{elem} ::= left-fringe | right-fringe | left-margin | right-margin
        |  scroll-bar | text
  @var{pos}  ::= left | center | right
  @var{form} ::= (@var{num} . @var{expr}) | (@var{op} @var{expr} ...)
  @var{op}   ::= + | -
@end group
@end smallexample

  The form @var{num} specifies a fraction of the default frame font
height or width.  The form @code{(@var{num})} specifies an absolute
number of pixels.  If @var{num} is a symbol, @var{symbol}, its
buffer-local variable binding is used.

  The @code{in}, @code{mm}, and @code{cm} units specify the number of
pixels per inch, millimeter, and centimeter, respectively.  The
@code{width} and @code{height} units correspond to the default width
and height of the current face.  An image specification @code{image}
corresponds to the width or height of the image.

  The elements @code{left-fringe}, @code{right-fringe},
@code{left-margin}, @code{right-margin}, @code{scroll-bar}, and
@code{text} specify to the width of the corresponding area of the
window.

  The @code{left}, @code{center}, and @code{right} positions can be
used with @code{:align-to} to specify a position relative to the left
edge, center, or right edge of the text area.

  Any of the above window elements (except @code{text}) can also be
used with @code{:align-to} to specify that the position is relative to
the left edge of the given area.  Once the base offset for a relative
position has been set (by the first occurrence of one of these
symbols), further occurrences of these symbols are interpreted as the
width of the specified area.  For example, to align to the center of
the left-margin, use

@example
:align-to (+ left-margin (0.5 . left-margin))
@end example

  If no specific base offset is set for alignment, it is always relative
to the left edge of the text area.  For example, @samp{:align-to 0} in a
header-line aligns with the first text column in the text area.

  A value of the form @code{(@var{num} . @var{expr})} stands for the
product of the values of @var{num} and @var{expr}.  For example,
@code{(2 . in)} specifies a width of 2 inches, while @code{(0.5 .
@var{image})} specifies half the width (or height) of the specified
image.

  The form @code{(+ @var{expr} ...)} adds up the value of the
expressions.  The form @code{(- @var{expr} ...)} negates or subtracts
the value of the expressions.

@node Other Display Specs
@subsection Other Display Specifications

  Here are the other sorts of display specifications that you can use
in the @code{display} text property.

@table @code
@item @var{string}
Display @var{string} instead of the text that has this property.

Recursive display specifications are not supported---@var{string}'s
@code{display} properties, if any, are not used.

@item (image . @var{image-props})
This kind of display specification is an image descriptor (@pxref{Images}).
When used as a display specification, it means to display the image
instead of the text that has the display specification.

@item (slice @var{x} @var{y} @var{width} @var{height})
This specification together with @code{image} specifies a @dfn{slice}
(a partial area) of the image to display.  The elements @var{y} and
@var{x} specify the top left corner of the slice, within the image;
@var{width} and @var{height} specify the width and height of the
slice.  Integer values are numbers of pixels.  A floating point number
in the range 0.0--1.0 stands for that fraction of the width or height
of the entire image.

@item ((margin nil) @var{string})
A display specification of this form means to display @var{string}
instead of the text that has the display specification, at the same
position as that text.  It is equivalent to using just @var{string},
but it is done as a special case of marginal display (@pxref{Display
Margins}).

@item (left-fringe @var{bitmap} @r{[}@var{face}@r{]})
@itemx (right-fringe @var{bitmap} @r{[}@var{face}@r{]})
This display specification on any character of a line of text causes
the specified @var{bitmap} be displayed in the left or right fringes
for that line, instead of the characters that have the display
specification.  The optional @var{face} specifies the colors to be
used for the bitmap.  @xref{Fringe Bitmaps}, for the details.

@item (space-width @var{factor})
This display specification affects all the space characters within the
text that has the specification.  It displays all of these spaces
@var{factor} times as wide as normal.  The element @var{factor} should
be an integer or float.  Characters other than spaces are not affected
at all; in particular, this has no effect on tab characters.

@item (height @var{height})
This display specification makes the text taller or shorter.
Here are the possibilities for @var{height}:

@table @asis
@item @code{(+ @var{n})}
@c FIXME: Add an index for "step"?  --xfq
This means to use a font that is @var{n} steps larger.  A ``step'' is
defined by the set of available fonts---specifically, those that match
what was otherwise specified for this text, in all attributes except
height.  Each size for which a suitable font is available counts as
another step.  @var{n} should be an integer.

@item @code{(- @var{n})}
This means to use a font that is @var{n} steps smaller.

@item a number, @var{factor}
A number, @var{factor}, means to use a font that is @var{factor} times
as tall as the default font.

@item a symbol, @var{function}
A symbol is a function to compute the height.  It is called with the
current height as argument, and should return the new height to use.

@item anything else, @var{form}
If the @var{height} value doesn't fit the previous possibilities, it is
a form.  Emacs evaluates it to get the new height, with the symbol
@code{height} bound to the current specified font height.
@end table

@item (raise @var{factor})
This kind of display specification raises or lowers the text
it applies to, relative to the baseline of the line.

@var{factor} must be a number, which is interpreted as a multiple of the
height of the affected text.  If it is positive, that means to display
the characters raised.  If it is negative, that means to display them
lower down.

If the text also has a @code{height} display specification, that does
not affect the amount of raising or lowering, which is based on the
faces used for the text.
@end table

@c We put all the `@code{(when ...)}' on one line to encourage
@c makeinfo's end-of-sentence heuristics to DTRT.  Previously, the dot
@c was at eol; the info file ended up w/ two spaces rendered after it.
  You can make any display specification conditional.  To do that,
package it in another list of the form
@code{(when @var{condition} . @var{spec})}.
Then the specification @var{spec} applies only when
@var{condition} evaluates to a non-@code{nil} value.  During the
evaluation, @code{object} is bound to the string or buffer having the
conditional @code{display} property.  @code{position} and
@code{buffer-position} are bound to the position within @code{object}
and the buffer position where the @code{display} property was found,
respectively.  Both positions can be different when @code{object} is a
string.

@node Display Margins
@subsection Displaying in the Margins
@cindex display margins
@cindex margins, display

  A buffer can have blank areas called @dfn{display margins} on the
left and on the right.  Ordinary text never appears in these areas,
but you can put things into the display margins using the
@code{display} property.  There is currently no way to make text or
images in the margin mouse-sensitive.

  The way to display something in the margins is to specify it in a
margin display specification in the @code{display} property of some
text.  This is a replacing display specification, meaning that the
text you put it on does not get displayed; the margin display appears,
but that text does not.

  A margin display specification looks like @code{((margin
right-margin) @var{spec})} or @code{((margin left-margin) @var{spec})}.
Here, @var{spec} is another display specification that says what to
display in the margin.  Typically it is a string of text to display,
or an image descriptor.

  To display something in the margin @emph{in association with}
certain buffer text, without altering or preventing the display of
that text, put a @code{before-string} property on the text and put the
margin display specification on the contents of the before-string.

  Before the display margins can display anything, you must give
them a nonzero width.  The usual way to do that is to set these
variables:

@defvar left-margin-width
This variable specifies the width of the left margin, in character
cell (a.k.a.@: ``column'') units.  It is buffer-local in all buffers.
A value of @code{nil} means no left marginal area.
@end defvar

@defvar right-margin-width
This variable specifies the width of the right margin, in character
cell units.  It is buffer-local in all buffers.  A value of @code{nil}
means no right marginal area.
@end defvar

  Setting these variables does not immediately affect the window.  These
variables are checked when a new buffer is displayed in the window.
Thus, you can make changes take effect by calling
@code{set-window-buffer}.

  You can also set the margin widths immediately.

@defun set-window-margins window left &optional right
This function specifies the margin widths for window @var{window}, in
character cell units.  The argument @var{left} controls the left
margin, and @var{right} controls the right margin (default @code{0}).
@end defun

@defun window-margins &optional window
This function returns the width of the left and right margins of
@var{window} as a cons cell of the form @w{@code{(@var{left}
. @var{right})}}.  If one of the two marginal areas does not exist,
its width is returned as @code{nil}; if neither of the two margins exist,
the function returns @code{(nil)}.  If @var{window} is @code{nil}, the
selected window is used.
@end defun

@node Images
@section Images
@cindex images in buffers

  To display an image in an Emacs buffer, you must first create an image
descriptor, then use it as a display specifier in the @code{display}
property of text that is displayed (@pxref{Display Property}).

  Emacs is usually able to display images when it is run on a
graphical terminal.  Images cannot be displayed in a text terminal, on
certain graphical terminals that lack the support for this, or if
Emacs is compiled without image support.  You can use the function
@code{display-images-p} to determine if images can in principle be
displayed (@pxref{Display Feature Testing}).

@menu
* Image Formats::       Supported image formats.
* Image Descriptors::   How to specify an image for use in @code{:display}.
* XBM Images::          Special features for XBM format.
* XPM Images::          Special features for XPM format.
* PostScript Images::   Special features for PostScript format.
* ImageMagick Images::  Special features available through ImageMagick.
* Other Image Types::   Various other formats are supported.
* Defining Images::     Convenient ways to define an image for later use.
* Showing Images::      Convenient ways to display an image once it is defined.
* Multi-Frame Images::  Some images contain more than one frame.
* Image Cache::         Internal mechanisms of image display.
@end menu

@node Image Formats
@subsection Image Formats
@cindex image formats
@cindex image types

  Emacs can display a number of different image formats.  Some of
these image formats are supported only if particular support libraries
are installed.  On some platforms, Emacs can load support libraries on
demand; if so, the variable @code{dynamic-library-alist} can be used
to modify the set of known names for these dynamic libraries.
@xref{Dynamic Libraries}.

  Supported image formats (and the required support libraries) include
PBM and XBM (which do not depend on support libraries and are always
available), XPM (@code{libXpm}), GIF (@code{libgif} or
@code{libungif}), PostScript (@code{gs}), JPEG (@code{libjpeg}), TIFF
(@code{libtiff}), PNG (@code{libpng}), and SVG (@code{librsvg}).

  Each of these image formats is associated with an @dfn{image type
symbol}.  The symbols for the above formats are, respectively,
@code{pbm}, @code{xbm}, @code{xpm}, @code{gif}, @code{postscript},
@code{jpeg}, @code{tiff}, @code{png}, and @code{svg}.

  Furthermore, if you build Emacs with ImageMagick
(@code{libMagickWand}) support, Emacs can display any image format
that ImageMagick can.  @xref{ImageMagick Images}.  All images
displayed via ImageMagick have type symbol @code{imagemagick}.

@defvar image-types
This variable contains a list of type symbols for image formats which
are potentially supported in the current configuration.

``Potentially'' means that Emacs knows about the image types, not
necessarily that they can be used (for example, they could depend on
unavailable dynamic libraries).  To know which image types are really
available, use @code{image-type-available-p}.
@end defvar

@defun image-type-available-p type
This function returns non-@code{nil} if images of type @var{type} can
be loaded and displayed.  @var{type} must be an image type symbol.

For image types whose support libraries are statically linked, this
function always returns @code{t}.  For image types whose support
libraries are dynamically loaded, it returns @code{t} if the library
could be loaded and @code{nil} otherwise.
@end defun

@node Image Descriptors
@subsection Image Descriptors
@cindex image descriptor

  An @dfn{image descriptor} is a list which specifies the underlying
data for an image, and how to display it.  It is typically used as the
value of a @code{display} overlay or text property (@pxref{Other
Display Specs}); but @xref{Showing Images}, for convenient helper
functions to insert images into buffers.

  Each image descriptor has the form @code{(image . @var{props})},
where @var{props} is a property list of alternating keyword symbols
and values, including at least the pair @code{:type @var{TYPE}} which
specifies the image type.

  The following is a list of properties that are meaningful for all
image types (there are also properties which are meaningful only for
certain image types, as documented in the following subsections):

@table @code
@item :type @var{type}
The image type.
@ifnottex
@xref{Image Formats}.
@end ifnottex
Every image descriptor must include this property.

@item :file @var{file}
This says to load the image from file @var{file}.  If @var{file} is
not an absolute file name, it is expanded in @code{data-directory}.

@item :data @var{data}
This specifies the raw image data.  Each image descriptor must have
either @code{:data} or @code{:file}, but not both.

For most image types, the value of a @code{:data} property should be a
string containing the image data.  Some image types do not support
@code{:data}; for some others, @code{:data} alone is not enough, so
you need to use other image properties along with @code{:data}.  See
the following subsections for details.

@item :margin @var{margin}
This specifies how many pixels to add as an extra margin around the
image.  The value, @var{margin}, must be a non-negative number, or a
pair @code{(@var{x} . @var{y})} of such numbers.  If it is a pair,
@var{x} specifies how many pixels to add horizontally, and @var{y}
specifies how many pixels to add vertically.  If @code{:margin} is not
specified, the default is zero.

@item :ascent @var{ascent}
This specifies the amount of the image's height to use for its
ascent---that is, the part above the baseline.  The value,
@var{ascent}, must be a number in the range 0 to 100, or the symbol
@code{center}.

If @var{ascent} is a number, that percentage of the image's height is
used for its ascent.

If @var{ascent} is @code{center}, the image is vertically centered
around a centerline which would be the vertical centerline of text drawn
at the position of the image, in the manner specified by the text
properties and overlays that apply to the image.

If this property is omitted, it defaults to 50.

@item :relief @var{relief}
This adds a shadow rectangle around the image.  The value,
@var{relief}, specifies the width of the shadow lines, in pixels.  If
@var{relief} is negative, shadows are drawn so that the image appears
as a pressed button; otherwise, it appears as an unpressed button.

@item :conversion @var{algorithm}
This specifies a conversion algorithm that should be applied to the
image before it is displayed; the value, @var{algorithm}, specifies
which algorithm.

@table @code
@item laplace
@itemx emboss
Specifies the Laplace edge detection algorithm, which blurs out small
differences in color while highlighting larger differences.  People
sometimes consider this useful for displaying the image for a
``disabled'' button.

@item (edge-detection :matrix @var{matrix} :color-adjust @var{adjust})
@cindex edge detection, images
Specifies a general edge-detection algorithm.  @var{matrix} must be
either a nine-element list or a nine-element vector of numbers.  A pixel
at position @math{x/y} in the transformed image is computed from
original pixels around that position.  @var{matrix} specifies, for each
pixel in the neighborhood of @math{x/y}, a factor with which that pixel
will influence the transformed pixel; element @math{0} specifies the
factor for the pixel at @math{x-1/y-1}, element @math{1} the factor for
the pixel at @math{x/y-1} etc., as shown below:
@iftex
@tex
$$\pmatrix{x-1/y-1 & x/y-1  & x+1/y-1 \cr
   x-1/y  &   x/y &    x+1/y \cr
   x-1/y+1&   x/y+1 &  x+1/y+1 \cr}$$
@end tex
@end iftex
@ifnottex
@display
  (x-1/y-1  x/y-1  x+1/y-1
   x-1/y    x/y    x+1/y
   x-1/y+1  x/y+1  x+1/y+1)
@end display
@end ifnottex

The resulting pixel is computed from the color intensity of the color
resulting from summing up the RGB values of surrounding pixels,
multiplied by the specified factors, and dividing that sum by the sum
of the factors' absolute values.

Laplace edge-detection currently uses a matrix of
@iftex
@tex
$$\pmatrix{1 & 0 & 0 \cr
   0&  0 &  0 \cr
   0 & 0 & -1 \cr}$$
@end tex
@end iftex
@ifnottex
@display
  (1  0  0
   0  0  0
   0  0 -1)
@end display
@end ifnottex

Emboss edge-detection uses a matrix of
@iftex
@tex
$$\pmatrix{ 2 & -1 &  0 \cr
   -1 &  0 &  1 \cr
    0  & 1 & -2 \cr}$$
@end tex
@end iftex
@ifnottex
@display
  ( 2 -1  0
   -1  0  1
    0  1 -2)
@end display
@end ifnottex

@item disabled
Specifies transforming the image so that it looks ``disabled''.
@end table

@item :mask @var{mask}
If @var{mask} is @code{heuristic} or @code{(heuristic @var{bg})}, build
a clipping mask for the image, so that the background of a frame is
visible behind the image.  If @var{bg} is not specified, or if @var{bg}
is @code{t}, determine the background color of the image by looking at
the four corners of the image, assuming the most frequently occurring
color from the corners is the background color of the image.  Otherwise,
@var{bg} must be a list @code{(@var{red} @var{green} @var{blue})}
specifying the color to assume for the background of the image.

If @var{mask} is @code{nil}, remove a mask from the image, if it has
one.  Images in some formats include a mask which can be removed by
specifying @code{:mask nil}.

@item :pointer @var{shape}
This specifies the pointer shape when the mouse pointer is over this
image.  @xref{Pointer Shape}, for available pointer shapes.

@item :map @var{map}
@cindex image maps
This associates an image map of @dfn{hot spots} with this image.

An image map is an alist where each element has the format
@code{(@var{area} @var{id} @var{plist})}.  An @var{area} is specified
as either a rectangle, a circle, or a polygon.

A rectangle is a cons
@code{(rect . ((@var{x0} . @var{y0}) . (@var{x1} . @var{y1})))}
which specifies the pixel coordinates of the upper left and bottom right
corners of the rectangle area.

A circle is a cons
@code{(circle . ((@var{x0} . @var{y0}) . @var{r}))}
which specifies the center and the radius of the circle; @var{r} may
be a float or integer.

A polygon is a cons
@code{(poly . [@var{x0} @var{y0} @var{x1} @var{y1} ...])}
where each pair in the vector describes one corner in the polygon.

When the mouse pointer lies on a hot-spot area of an image, the
@var{plist} of that hot-spot is consulted; if it contains a @code{help-echo}
property, that defines a tool-tip for the hot-spot, and if it contains
a @code{pointer} property, that defines the shape of the mouse cursor when
it is on the hot-spot.
@xref{Pointer Shape}, for available pointer shapes.

When you click the mouse when the mouse pointer is over a hot-spot, an
event is composed by combining the @var{id} of the hot-spot with the
mouse event; for instance, @code{[area4 mouse-1]} if the hot-spot's
@var{id} is @code{area4}.
@end table

@defun image-mask-p spec &optional frame
This function returns @code{t} if image @var{spec} has a mask bitmap.
@var{frame} is the frame on which the image will be displayed.
@var{frame} @code{nil} or omitted means to use the selected frame
(@pxref{Input Focus}).
@end defun

@node XBM Images
@subsection XBM Images
@cindex XBM

  To use XBM format, specify @code{xbm} as the image type.  This image
format doesn't require an external library, so images of this type are
always supported.

  Additional image properties supported for the @code{xbm} image type are:

@table @code
@item :foreground @var{foreground}
The value, @var{foreground}, should be a string specifying the image
foreground color, or @code{nil} for the default color.  This color is
used for each pixel in the XBM that is 1.  The default is the frame's
foreground color.

@item :background @var{background}
The value, @var{background}, should be a string specifying the image
background color, or @code{nil} for the default color.  This color is
used for each pixel in the XBM that is 0.  The default is the frame's
background color.
@end table

  If you specify an XBM image using data within Emacs instead of an
external file, use the following three properties:

@table @code
@item :data @var{data}
The value, @var{data}, specifies the contents of the image.
There are three formats you can use for @var{data}:

@itemize @bullet
@item
A vector of strings or bool-vectors, each specifying one line of the
image.  Do specify @code{:height} and @code{:width}.

@item
A string containing the same byte sequence as an XBM file would contain.
You must not specify @code{:height} and @code{:width} in this case,
because omitting them is what indicates the data has the format of an
XBM file.  The file contents specify the height and width of the image.

@item
A string or a bool-vector containing the bits of the image (plus perhaps
some extra bits at the end that will not be used).  It should contain at
least @var{width} * @code{height} bits.  In this case, you must specify
@code{:height} and @code{:width}, both to indicate that the string
contains just the bits rather than a whole XBM file, and to specify the
size of the image.
@end itemize

@item :width @var{width}
The value, @var{width}, specifies the width of the image, in pixels.

@item :height @var{height}
The value, @var{height}, specifies the height of the image, in pixels.
@end table

@node XPM Images
@subsection XPM Images
@cindex XPM

  To use XPM format, specify @code{xpm} as the image type.  The
additional image property @code{:color-symbols} is also meaningful with
the @code{xpm} image type:

@table @code
@item :color-symbols @var{symbols}
The value, @var{symbols}, should be an alist whose elements have the
form @code{(@var{name} . @var{color})}.  In each element, @var{name} is
the name of a color as it appears in the image file, and @var{color}
specifies the actual color to use for displaying that name.
@end table

@node PostScript Images
@subsection PostScript Images
@cindex postscript images

  To use PostScript for an image, specify image type @code{postscript}.
This works only if you have Ghostscript installed.  You must always use
these three properties:

@table @code
@item :pt-width @var{width}
The value, @var{width}, specifies the width of the image measured in
points (1/72 inch).  @var{width} must be an integer.

@item :pt-height @var{height}
The value, @var{height}, specifies the height of the image in points
(1/72 inch).  @var{height} must be an integer.

@item :bounding-box @var{box}
The value, @var{box}, must be a list or vector of four integers, which
specifying the bounding box of the PostScript image, analogous to the
@samp{BoundingBox} comment found in PostScript files.

@example
%%BoundingBox: 22 171 567 738
@end example
@end table

@node ImageMagick Images
@subsection ImageMagick Images
@cindex ImageMagick images
@cindex images, support for more formats

  If you build Emacs with ImageMagick support, you can use the
ImageMagick library to load many image formats (@pxref{File
Conveniences,,, emacs, The GNU Emacs Manual}).  The image type symbol
for images loaded via ImageMagick is @code{imagemagick}, regardless of
the actual underlying image format.

@defun imagemagick-types
This function returns a list of image file extensions supported by the
current ImageMagick installation.  Each list element is a symbol
representing an internal ImageMagick name for an image type, such as
@code{BMP} for @file{.bmp} images.
@end defun

@defopt imagemagick-enabled-types
The value of this variable is a list of ImageMagick image types which
Emacs may attempt to render using ImageMagick.  Each list element
should be one of the symbols in the list returned by
@code{imagemagick-types}, or an equivalent string.  Alternatively, a
value of @code{t} enables ImageMagick for all possible image types.
Regardless of the value of this variable,
@code{imagemagick-types-inhibit} (see below) takes precedence.
@end defopt

@defopt imagemagick-types-inhibit
The value of this variable lists the ImageMagick image types which
should never be rendered using ImageMagick, regardless of the value of
@code{imagemagick-enabled-types}.  A value of @code{t} disables
ImageMagick entirely.
@end defopt

@defvar image-format-suffixes
This variable is an alist mapping image types to file name extensions.
Emacs uses this in conjunction with the @code{:format} image property
(see below) to give a hint to the ImageMagick library as to the type
of an image.  Each element has the form @code{(@var{type}
@var{extension})}, where @var{type} is a symbol specifying an image
content-type, and @var{extension} is a string that specifies the
associated file name extension.
@end defvar

  Images loaded with ImageMagick support the following additional
image descriptor properties:

@table @code
@item :background @var{background}
@var{background}, if non-@code{nil}, should be a string specifying a
color, which is used as the image's background color if the image
supports transparency.  If the value is @code{nil}, it defaults to the
frame's background color.

@item :width @var{width}, :height @var{height}
The @code{:width} and @code{:height} keywords are used for scaling the
image.  If only one of them is specified, the other one will be
calculated so as to preserve the aspect ratio.  If both are specified,
aspect ratio may not be preserved.

@item :max-width @var{max-width}, :max-height @var{max-height}
The @code{:max-width} and @code{:max-height} keywords are used for
scaling if the size of the image of the image exceeds these values.
If @code{:width} is set it will have precedence over @code{max-width},
and if @code{:height} is set it will have precedence over
@code{max-height}, but you can otherwise mix these keywords as you
wish.  @code{:max-width} and @code{:max-height} will always preserve
the aspect ratio.

@item :format @var{type}
The value, @var{type}, should be a symbol specifying the type of the
image data, as found in @code{image-format-suffixes}.  This is used
when the image does not have an associated file name, to provide a
hint to ImageMagick to help it detect the image type.

@item :rotation @var{angle}
Specifies a rotation angle in degrees.

@item :index @var{frame}
@c Doesn't work: http://debbugs.gnu.org/7978
@xref{Multi-Frame Images}.
@end table

@node Other Image Types
@subsection Other Image Types
@cindex PBM

  For PBM images, specify image type @code{pbm}.  Color, gray-scale and
monochromatic images are supported.   For mono PBM images, two additional
image properties are supported.

@table @code
@item :foreground @var{foreground}
The value, @var{foreground}, should be a string specifying the image
foreground color, or @code{nil} for the default color.  This color is
used for each pixel in the PBM that is 1.  The default is the frame's
foreground color.

@item :background @var{background}
The value, @var{background}, should be a string specifying the image
background color, or @code{nil} for the default color.  This color is
used for each pixel in the PBM that is 0.  The default is the frame's
background color.
@end table

@noindent
The remaining image types that Emacs can support are:

@table @asis
@item GIF
Image type @code{gif}.
Supports the @code{:index} property.  @xref{Multi-Frame Images}.

@item JPEG
Image type @code{jpeg}.

@item PNG
Image type @code{png}.

@item SVG
Image type @code{svg}.

@item TIFF
Image type @code{tiff}.
Supports the @code{:index} property.  @xref{Multi-Frame Images}.
@end table

@node Defining Images
@subsection Defining Images

  The functions @code{create-image}, @code{defimage} and
@code{find-image} provide convenient ways to create image descriptors.

@defun create-image file-or-data &optional type data-p &rest props
This function creates and returns an image descriptor which uses the
data in @var{file-or-data}.  @var{file-or-data} can be a file name or
a string containing the image data; @var{data-p} should be @code{nil}
for the former case, non-@code{nil} for the latter case.

The optional argument @var{type} is a symbol specifying the image type.
If @var{type} is omitted or @code{nil}, @code{create-image} tries to
determine the image type from the file's first few bytes, or else
from the file's name.

The remaining arguments, @var{props}, specify additional image
properties---for example,

@c ':heuristic-mask' is not documented?
@example
(create-image "foo.xpm" 'xpm nil :heuristic-mask t)
@end example

The function returns @code{nil} if images of this type are not
supported.  Otherwise it returns an image descriptor.
@end defun

@defmac defimage symbol specs &optional doc
This macro defines @var{symbol} as an image name.  The arguments
@var{specs} is a list which specifies how to display the image.
The third argument, @var{doc}, is an optional documentation string.

Each argument in @var{specs} has the form of a property list, and each
one should specify at least the @code{:type} property and either the
@code{:file} or the @code{:data} property.  The value of @code{:type}
should be a symbol specifying the image type, the value of
@code{:file} is the file to load the image from, and the value of
@code{:data} is a string containing the actual image data.  Here is an
example:

@example
(defimage test-image
  ((:type xpm :file "~/test1.xpm")
   (:type xbm :file "~/test1.xbm")))
@end example

@code{defimage} tests each argument, one by one, to see if it is
usable---that is, if the type is supported and the file exists.  The
first usable argument is used to make an image descriptor which is
stored in @var{symbol}.

If none of the alternatives will work, then @var{symbol} is defined
as @code{nil}.
@end defmac

@defun find-image specs
This function provides a convenient way to find an image satisfying one
of a list of image specifications @var{specs}.

Each specification in @var{specs} is a property list with contents
depending on image type.  All specifications must at least contain the
properties @code{:type @var{type}} and either @w{@code{:file @var{file}}}
or @w{@code{:data @var{DATA}}}, where @var{type} is a symbol specifying
the image type, e.g., @code{xbm}, @var{file} is the file to load the
image from, and @var{data} is a string containing the actual image data.
The first specification in the list whose @var{type} is supported, and
@var{file} exists, is used to construct the image specification to be
returned.  If no specification is satisfied, @code{nil} is returned.

The image is looked for in @code{image-load-path}.
@end defun

@defvar image-load-path
This variable's value is a list of locations in which to search for
image files.  If an element is a string or a variable symbol whose
value is a string, the string is taken to be the name of a directory
to search.  If an element is a variable symbol whose value is a list,
that is taken to be a list of directory names to search.

The default is to search in the @file{images} subdirectory of the
directory specified by @code{data-directory}, then the directory
specified by @code{data-directory}, and finally in the directories in
@code{load-path}.  Subdirectories are not automatically included in
the search, so if you put an image file in a subdirectory, you have to
supply the subdirectory name explicitly.  For example, to find the
image @file{images/foo/bar.xpm} within @code{data-directory}, you
should specify the image as follows:

@example
(defimage foo-image '((:type xpm :file "foo/bar.xpm")))
@end example
@end defvar

@defun image-load-path-for-library library image &optional path no-error
This function returns a suitable search path for images used by the
Lisp package @var{library}.

The function searches for @var{image} first using @code{image-load-path},
excluding @file{@code{data-directory}/images}, and then in
@code{load-path}, followed by a path suitable for @var{library}, which
includes @file{../../etc/images} and @file{../etc/images} relative to
the library file itself, and finally in
@file{@code{data-directory}/images}.

Then this function returns a list of directories which contains first
the directory in which @var{image} was found, followed by the value of
@code{load-path}.  If @var{path} is given, it is used instead of
@code{load-path}.

If @var{no-error} is non-@code{nil} and a suitable path can't be
found, don't signal an error.  Instead, return a list of directories as
before, except that @code{nil} appears in place of the image directory.

Here is an example of using @code{image-load-path-for-library}:

@example
(defvar image-load-path) ; shush compiler
(let* ((load-path (image-load-path-for-library
                    "mh-e" "mh-logo.xpm"))
       (image-load-path (cons (car load-path)
                              image-load-path)))
  (mh-tool-bar-folder-buttons-init))
@end example
@end defun

@node Showing Images
@subsection Showing Images

  You can use an image descriptor by setting up the @code{display}
property yourself, but it is easier to use the functions in this
section.

@defun insert-image image &optional string area slice
This function inserts @var{image} in the current buffer at point.  The
value @var{image} should be an image descriptor; it could be a value
returned by @code{create-image}, or the value of a symbol defined with
@code{defimage}.  The argument @var{string} specifies the text to put
in the buffer to hold the image.  If it is omitted or @code{nil},
@code{insert-image} uses @code{" "} by default.

The argument @var{area} specifies whether to put the image in a margin.
If it is @code{left-margin}, the image appears in the left margin;
@code{right-margin} specifies the right margin.  If @var{area} is
@code{nil} or omitted, the image is displayed at point within the
buffer's text.

The argument @var{slice} specifies a slice of the image to insert.  If
@var{slice} is @code{nil} or omitted the whole image is inserted.
Otherwise, @var{slice} is a list @code{(@var{x} @var{y} @var{width}
@var{height})} which specifies the @var{x} and @var{y} positions and
@var{width} and @var{height} of the image area to insert.  Integer
values are in units of pixels.  A floating point number in the range
0.0--1.0 stands for that fraction of the width or height of the entire
image.

Internally, this function inserts @var{string} in the buffer, and gives
it a @code{display} property which specifies @var{image}.  @xref{Display
Property}.
@end defun

@cindex slice, image
@cindex image slice
@defun insert-sliced-image image &optional string area rows cols
This function inserts @var{image} in the current buffer at point, like
@code{insert-image}, but splits the image into @var{rows}x@var{cols}
equally sized slices.

If an image is inserted ``sliced'', Emacs displays each slice as a
separate image, and allow more intuitive scrolling up/down, instead of
jumping up/down the entire image when paging through a buffer that
displays (large) images.
@end defun

@defun put-image image pos &optional string area
This function puts image @var{image} in front of @var{pos} in the
current buffer.  The argument @var{pos} should be an integer or a
marker.  It specifies the buffer position where the image should appear.
The argument @var{string} specifies the text that should hold the image
as an alternative to the default.

The argument @var{image} must be an image descriptor, perhaps returned
by @code{create-image} or stored by @code{defimage}.

The argument @var{area} specifies whether to put the image in a margin.
If it is @code{left-margin}, the image appears in the left margin;
@code{right-margin} specifies the right margin.  If @var{area} is
@code{nil} or omitted, the image is displayed at point within the
buffer's text.

Internally, this function creates an overlay, and gives it a
@code{before-string} property containing text that has a @code{display}
property whose value is the image.  (Whew!)
@end defun

@defun remove-images start end &optional buffer
This function removes images in @var{buffer} between positions
@var{start} and @var{end}.  If @var{buffer} is omitted or @code{nil},
images are removed from the current buffer.

This removes only images that were put into @var{buffer} the way
@code{put-image} does it, not images that were inserted with
@code{insert-image} or in other ways.
@end defun

@defun image-size spec &optional pixels frame
@cindex size of image
This function returns the size of an image as a pair
@w{@code{(@var{width} . @var{height})}}.  @var{spec} is an image
specification.  @var{pixels} non-@code{nil} means return sizes
measured in pixels, otherwise return sizes measured in canonical
character units (fractions of the width/height of the frame's default
font).  @var{frame} is the frame on which the image will be displayed.
@var{frame} null or omitted means use the selected frame (@pxref{Input
Focus}).
@end defun

@defvar max-image-size
This variable is used to define the maximum size of image that Emacs
will load.  Emacs will refuse to load (and display) any image that is
larger than this limit.

If the value is an integer, it directly specifies the maximum
image height and width, measured in pixels.  If it is a floating
point number, it specifies the maximum image height and width
as a ratio to the frame height and width.  If the value is
non-numeric, there is no explicit limit on the size of images.

The purpose of this variable is to prevent unreasonably large images
from accidentally being loaded into Emacs.  It only takes effect the
first time an image is loaded.  Once an image is placed in the image
cache, it can always be displayed, even if the value of
@code{max-image-size} is subsequently changed (@pxref{Image Cache}).
@end defvar

@node Multi-Frame Images
@subsection Multi-Frame Images
@cindex multi-frame images

@cindex animation
@cindex image animation
@cindex image frames
Some image files can contain more than one image.  We say that there
are multiple ``frames'' in the image.  At present, Emacs supports
multiple frames for GIF, TIFF, and certain ImageMagick formats such as
DJVM@.

The frames can be used either to represent multiple ``pages'' (this is
usually the case with multi-frame TIFF files, for example), or to
create animation (usually the case with multi-frame GIF files).

A multi-frame image has a property @code{:index}, whose value is an
integer (counting from 0) that specifies which frame is being displayed.

@defun image-multi-frame-p image
This function returns non-@code{nil} if @var{image} contains more than
one frame.  The actual return value is a cons @code{(@var{nimages}
. @var{delay})}, where @var{nimages} is the number of frames and
@var{delay} is the delay in seconds between them, or @code{nil}
if the image does not specify a delay.  Images that are intended to be
animated usually specify a frame delay, whereas ones that are intended
to be treated as multiple pages do not.
@end defun

@defun image-current-frame image
This function returns the index of the current frame number for
@var{image}, counting from 0.
@end defun

@defun image-show-frame image n &optional nocheck
This function switches @var{image} to frame number @var{n}.  It
replaces a frame number outside the valid range with that of the end
of the range, unless @var{nocheck} is non-@code{nil}.  If @var{image}
does not contain a frame with the specified number, the image displays
as a hollow box.
@end defun

@defun image-animate image &optional index limit
This function animates @var{image}.  The optional integer @var{index}
specifies the frame from which to start (default 0).  The optional
argument @var{limit} controls the length of the animation.  If omitted
or @code{nil}, the image animates once only; if @code{t} it loops
forever; if a number animation stops after that many seconds.
@end defun

@vindex image-minimum-frame-delay
@vindex image-default-frame-delay
@noindent Animation operates by means of a timer.  Note that Emacs imposes a
minimum frame delay of 0.01 (@code{image-minimum-frame-delay}) seconds.
If the image itself does not specify a delay, Emacs uses
@code{image-default-frame-delay}.

@defun image-animate-timer image
This function returns the timer responsible for animating @var{image},
if there is one.
@end defun


@node Image Cache
@subsection Image Cache
@cindex image cache

  Emacs caches images so that it can display them again more
efficiently.  When Emacs displays an image, it searches the image
cache for an existing image specification @code{equal} to the desired
specification.  If a match is found, the image is displayed from the
cache.  Otherwise, Emacs loads the image normally.

@defun image-flush spec &optional frame
This function removes the image with specification @var{spec} from the
image cache of frame @var{frame}.  Image specifications are compared
using @code{equal}.  If @var{frame} is @code{nil}, it defaults to the
selected frame.  If @var{frame} is @code{t}, the image is flushed on
all existing frames.

In Emacs's current implementation, each graphical terminal possesses an
image cache, which is shared by all the frames on that terminal
(@pxref{Multiple Terminals}).  Thus, refreshing an image in one frame
also refreshes it in all other frames on the same terminal.
@end defun

  One use for @code{image-flush} is to tell Emacs about a change in an
image file.  If an image specification contains a @code{:file}
property, the image is cached based on the file's contents when the
image is first displayed.  Even if the file subsequently changes,
Emacs continues displaying the old version of the image.  Calling
@code{image-flush} flushes the image from the cache, forcing Emacs to
re-read the file the next time it needs to display that image.

  Another use for @code{image-flush} is for memory conservation.  If
your Lisp program creates a large number of temporary images over a
period much shorter than @code{image-cache-eviction-delay} (see
below), you can opt to flush unused images yourself, instead of
waiting for Emacs to do it automatically.

@defun clear-image-cache &optional filter
This function clears an image cache, removing all the images stored in
it.  If @var{filter} is omitted or @code{nil}, it clears the cache for
the selected frame.  If @var{filter} is a frame, it clears the cache
for that frame.  If @var{filter} is @code{t}, all image caches are
cleared.  Otherwise, @var{filter} is taken to be a file name, and all
images associated with that file name are removed from all image
caches.
@end defun

If an image in the image cache has not been displayed for a specified
period of time, Emacs removes it from the cache and frees the
associated memory.

@defvar image-cache-eviction-delay
This variable specifies the number of seconds an image can remain in
the cache without being displayed.  When an image is not displayed for
this length of time, Emacs removes it from the image cache.

Under some circumstances, if the number of images in the cache grows
too large, the actual eviction delay may be shorter than this.

If the value is @code{nil}, Emacs does not remove images from the cache
except when you explicitly clear it.  This mode can be useful for
debugging.
@end defvar

@node Buttons
@section Buttons
@cindex buttons in buffers
@cindex clickable buttons in buffers

  The Button package defines functions for inserting and manipulating
@dfn{buttons} that can be activated with the mouse or via keyboard
commands.  These buttons are typically used for various kinds of
hyperlinks.

  A button is essentially a set of text or overlay properties,
attached to a stretch of text in a buffer.  These properties are
called @dfn{button properties}.  One of these properties, the
@dfn{action property}, specifies a function which is called when the
user invokes the button using the keyboard or the mouse.  The action
function may examine the button and use its other properties as
desired.

  In some ways, the Button package duplicates the functionality in the
Widget package.  @xref{Top, , Introduction, widget, The Emacs Widget
Library}.  The advantage of the Button package is that it is faster,
smaller, and simpler to program.  From the point of view of the user,
the interfaces produced by the two packages are very similar.

@menu
* Button Properties::      Button properties with special meanings.
* Button Types::           Defining common properties for classes of buttons.
* Making Buttons::         Adding buttons to Emacs buffers.
* Manipulating Buttons::   Getting and setting properties of buttons.
* Button Buffer Commands:: Buffer-wide commands and bindings for buttons.
@end menu

@node Button Properties
@subsection Button Properties
@cindex button properties

  Each button has an associated list of properties defining its
appearance and behavior, and other arbitrary properties may be used
for application specific purposes.  The following properties have
special meaning to the Button package:

@table @code
@item action
@kindex action @r{(button property)}
The function to call when the user invokes the button, which is passed
the single argument @var{button}.  By default this is @code{ignore},
which does nothing.

@item mouse-action
@kindex mouse-action @r{(button property)}
This is similar to @code{action}, and when present, will be used
instead of @code{action} for button invocations resulting from
mouse-clicks (instead of the user hitting @key{RET}).  If not
present, mouse-clicks use @code{action} instead.

@item face
@kindex face @r{(button property)}
This is an Emacs face controlling how buttons of this type are
displayed; by default this is the @code{button} face.

@item mouse-face
@kindex mouse-face @r{(button property)}
This is an additional face which controls appearance during
mouse-overs (merged with the usual button face); by default this is
the usual Emacs @code{highlight} face.

@item keymap
@kindex keymap @r{(button property)}
The button's keymap, defining bindings active within the button
region.  By default this is the usual button region keymap, stored
in the variable @code{button-map}, which defines @key{RET} and
@key{mouse-2} to invoke the button.

@item type
@kindex type @r{(button property)}
The button type.  @xref{Button Types}.

@item help-echo
@kindex help-index @r{(button property)}
A string displayed by the Emacs tool-tip help system; by default,
@code{"mouse-2, RET: Push this button"}.

@item follow-link
@kindex follow-link @r{(button property)}
The follow-link property, defining how a @key{Mouse-1} click behaves
on this button, @xref{Clickable Text}.

@item button
@kindex button @r{(button property)}
All buttons have a non-@code{nil} @code{button} property, which may be useful
in finding regions of text that comprise buttons (which is what the
standard button functions do).
@end table

  There are other properties defined for the regions of text in a
button, but these are not generally interesting for typical uses.

@node Button Types
@subsection Button Types
@cindex button types

  Every button has a @dfn{button type}, which defines default values
for the button's properties.  Button types are arranged in a
hierarchy, with specialized types inheriting from more general types,
so that it's easy to define special-purpose types of buttons for
specific tasks.

@defun define-button-type name &rest properties
Define a `button type' called @var{name} (a symbol).
The remaining arguments
form a sequence of @var{property value} pairs, specifying default
property values for buttons with this type (a button's type may be set
by giving it a @code{type} property when creating the button, using
the @code{:type} keyword argument).

In addition, the keyword argument @code{:supertype} may be used to
specify a button-type from which @var{name} inherits its default
property values.  Note that this inheritance happens only when
@var{name} is defined; subsequent changes to a supertype are not
reflected in its subtypes.
@end defun

  Using @code{define-button-type} to define default properties for
buttons is not necessary---buttons without any specified type use the
built-in button-type @code{button}---but it is encouraged, since
doing so usually makes the resulting code clearer and more efficient.

@node Making Buttons
@subsection Making Buttons
@cindex making buttons

  Buttons are associated with a region of text, using an overlay or
text properties to hold button-specific information, all of which are
initialized from the button's type (which defaults to the built-in
button type @code{button}).  Like all Emacs text, the appearance of
the button is governed by the @code{face} property; by default (via
the @code{face} property inherited from the @code{button} button-type)
this is a simple underline, like a typical web-page link.

  For convenience, there are two sorts of button-creation functions,
those that add button properties to an existing region of a buffer,
called @code{make-...button}, and those that also insert the button
text, called @code{insert-...button}.

  The button-creation functions all take the @code{&rest} argument
@var{properties}, which should be a sequence of @var{property value}
pairs, specifying properties to add to the button; see @ref{Button
Properties}.  In addition, the keyword argument @code{:type} may be
used to specify a button-type from which to inherit other properties;
see @ref{Button Types}.  Any properties not explicitly specified
during creation will be inherited from the button's type (if the type
defines such a property).

  The following functions add a button using an overlay
(@pxref{Overlays}) to hold the button properties:

@defun make-button beg end &rest properties
This makes a button from @var{beg} to @var{end} in the
current buffer, and returns it.
@end defun

@defun insert-button label &rest properties
This insert a button with the label @var{label} at point,
and returns it.
@end defun

  The following functions are similar, but using text properties
(@pxref{Text Properties}) to hold the button properties.  Such buttons
do not add markers to the buffer, so editing in the buffer does not
slow down if there is an extremely large numbers of buttons.  However,
if there is an existing face text property on the text (e.g., a face
assigned by Font Lock mode), the button face may not be visible.  Both
of these functions return the starting position of the new button.

@defun make-text-button beg end &rest properties
This makes a button from @var{beg} to @var{end} in the current buffer,
using text properties.
@end defun

@defun insert-text-button label &rest properties
This inserts a button with the label @var{label} at point, using text
properties.
@end defun

@node Manipulating Buttons
@subsection Manipulating Buttons
@cindex manipulating buttons

These are functions for getting and setting properties of buttons.
Often these are used by a button's invocation function to determine
what to do.

Where a @var{button} parameter is specified, it means an object
referring to a specific button, either an overlay (for overlay
buttons), or a buffer-position or marker (for text property buttons).
Such an object is passed as the first argument to a button's
invocation function when it is invoked.

@defun button-start button
Return the position at which @var{button} starts.
@end defun

@defun button-end button
Return the position at which @var{button} ends.
@end defun

@defun button-get button prop
Get the property of button @var{button} named @var{prop}.
@end defun

@defun button-put button prop val
Set @var{button}'s @var{prop} property to @var{val}.
@end defun

@defun button-activate button &optional use-mouse-action
Call @var{button}'s @code{action} property (i.e., invoke it).  If
@var{use-mouse-action} is non-@code{nil}, try to invoke the button's
@code{mouse-action} property instead of @code{action}; if the button
has no @code{mouse-action} property, use @code{action} as normal.
@end defun

@defun button-label button
Return @var{button}'s text label.
@end defun

@defun button-type button
Return @var{button}'s button-type.
@end defun

@defun button-has-type-p button type
Return @code{t} if @var{button} has button-type @var{type}, or one of
@var{type}'s subtypes.
@end defun

@defun button-at pos
Return the button at position @var{pos} in the current buffer, or
@code{nil}.  If the button at @var{pos} is a text property button, the
return value is a marker pointing to @var{pos}.
@end defun

@defun button-type-put type prop val
Set the button-type @var{type}'s @var{prop} property to @var{val}.
@end defun

@defun button-type-get type prop
Get the property of button-type @var{type} named @var{prop}.
@end defun

@defun button-type-subtype-p type supertype
Return @code{t} if button-type @var{type} is a subtype of @var{supertype}.
@end defun

@node Button Buffer Commands
@subsection Button Buffer Commands
@cindex button buffer commands

These are commands and functions for locating and operating on
buttons in an Emacs buffer.

@code{push-button} is the command that a user uses to actually `push'
a button, and is bound by default in the button itself to @key{RET}
and to @key{mouse-2} using a local keymap in the button's overlay or
text properties.  Commands that are useful outside the buttons itself,
such as @code{forward-button} and @code{backward-button} are
additionally available in the keymap stored in
@code{button-buffer-map}; a mode which uses buttons may want to use
@code{button-buffer-map} as a parent keymap for its keymap.

If the button has a non-@code{nil} @code{follow-link} property, and
@code{mouse-1-click-follows-link} is set, a quick @key{Mouse-1} click
will also activate the @code{push-button} command.
@xref{Clickable Text}.

@deffn Command push-button &optional pos use-mouse-action
Perform the action specified by a button at location @var{pos}.
@var{pos} may be either a buffer position or a mouse-event.  If
@var{use-mouse-action} is non-@code{nil}, or @var{pos} is a
mouse-event (@pxref{Mouse Events}), try to invoke the button's
@code{mouse-action} property instead of @code{action}; if the button
has no @code{mouse-action} property, use @code{action} as normal.
@var{pos} defaults to point, except when @code{push-button} is invoked
interactively as the result of a mouse-event, in which case, the mouse
event's position is used.  If there's no button at @var{pos}, do
nothing and return @code{nil}, otherwise return @code{t}.
@end deffn

@deffn Command forward-button n &optional wrap display-message
Move to the @var{n}th next button, or @var{n}th previous button if
@var{n} is negative.  If @var{n} is zero, move to the start of any
button at point.  If @var{wrap} is non-@code{nil}, moving past either
end of the buffer continues from the other end.  If
@var{display-message} is non-@code{nil}, the button's help-echo string
is displayed.  Any button with a non-@code{nil} @code{skip} property
is skipped over.  Returns the button found.
@end deffn

@deffn Command backward-button n &optional wrap display-message
Move to the @var{n}th previous button, or @var{n}th next button if
@var{n} is negative.  If @var{n} is zero, move to the start of any
button at point.  If @var{wrap} is non-@code{nil}, moving past either
end of the buffer continues from the other end.  If
@var{display-message} is non-@code{nil}, the button's help-echo string
is displayed.  Any button with a non-@code{nil} @code{skip} property
is skipped over.  Returns the button found.
@end deffn

@defun next-button pos &optional count-current
@defunx previous-button pos &optional count-current
Return the next button after (for @code{next-button}) or before (for
@code{previous-button}) position @var{pos} in the current buffer.  If
@var{count-current} is non-@code{nil}, count any button at @var{pos}
in the search, instead of starting at the next button.
@end defun

@node Abstract Display
@section Abstract Display
@cindex ewoc
@cindex display, abstract
@cindex display, arbitrary objects
@cindex model/view/controller
@cindex view part, model/view/controller

  The Ewoc package constructs buffer text that represents a structure
of Lisp objects, and updates the text to follow changes in that
structure.  This is like the ``view'' component in the
``model/view/controller'' design paradigm.  Ewoc means ``Emacs's
Widget for Object Collections''.

  An @dfn{ewoc} is a structure that organizes information required to
construct buffer text that represents certain Lisp data.  The buffer
text of the ewoc has three parts, in order: first, fixed @dfn{header}
text; next, textual descriptions of a series of data elements (Lisp
objects that you specify); and last, fixed @dfn{footer} text.
Specifically, an ewoc contains information on:

@itemize @bullet
@item
The buffer which its text is generated in.

@item
The text's start position in the buffer.

@item
The header and footer strings.

@item
@cindex node, ewoc
@c or "@cindex node, abstract display"?
A doubly-linked chain of @dfn{nodes}, each of which contains:

@itemize
@item
A @dfn{data element}, a single Lisp object.

@item
Links to the preceding and following nodes in the chain.
@end itemize

@item
A @dfn{pretty-printer} function which is responsible for
inserting the textual representation of a data
element value into the current buffer.
@end itemize

  Typically, you define an ewoc with @code{ewoc-create}, and then pass
the resulting ewoc structure to other functions in the Ewoc package to
build nodes within it, and display it in the buffer.  Once it is
displayed in the buffer, other functions determine the correspondence
between buffer positions and nodes, move point from one node's textual
representation to another, and so forth.  @xref{Abstract Display
Functions}.

@cindex encapsulation, ewoc
@c or "@cindex encapsulation, abstract display"?
  A node @dfn{encapsulates} a data element much the way a variable
holds a value.  Normally, encapsulation occurs as a part of adding a
node to the ewoc.  You can retrieve the data element value and place a
new value in its place, like so:

@lisp
(ewoc-data @var{node})
@result{} value

(ewoc-set-data @var{node} @var{new-value})
@result{} @var{new-value}
@end lisp

@noindent
You can also use, as the data element value, a Lisp object (list or
vector) that is a container for the ``real'' value, or an index into
some other structure.  The example (@pxref{Abstract Display Example})
uses the latter approach.

  When the data changes, you will want to update the text in the
buffer.  You can update all nodes by calling @code{ewoc-refresh}, or
just specific nodes using @code{ewoc-invalidate}, or all nodes
satisfying a predicate using @code{ewoc-map}.  Alternatively, you can
delete invalid nodes using @code{ewoc-delete} or @code{ewoc-filter},
and add new nodes in their place.  Deleting a node from an ewoc deletes
its associated textual description from buffer, as well.

@menu
* Abstract Display Functions::  Functions in the Ewoc package.
* Abstract Display Example::    Example of using Ewoc.
@end menu

@node Abstract Display Functions
@subsection Abstract Display Functions

  In this subsection, @var{ewoc} and @var{node} stand for the
structures described above (@pxref{Abstract Display}), while
@var{data} stands for an arbitrary Lisp object used as a data element.

@defun ewoc-create pretty-printer &optional header footer nosep
This constructs and returns a new ewoc, with no nodes (and thus no data
elements).  @var{pretty-printer} should be a function that takes one
argument, a data element of the sort you plan to use in this ewoc, and
inserts its textual description at point using @code{insert} (and never
@code{insert-before-markers}, because that would interfere with the
Ewoc package's internal mechanisms).

Normally, a newline is automatically inserted after the header,
the footer and every node's textual description.  If @var{nosep}
is non-@code{nil}, no newline is inserted.  This may be useful for
displaying an entire ewoc on a single line, for example, or for
making nodes ``invisible'' by arranging for @var{pretty-printer}
to do nothing for those nodes.

An ewoc maintains its text in the buffer that is current when
you create it, so switch to the intended buffer before calling
@code{ewoc-create}.
@end defun

@defun ewoc-buffer ewoc
This returns the buffer where @var{ewoc} maintains its text.
@end defun

@defun ewoc-get-hf ewoc
This returns a cons cell @code{(@var{header} . @var{footer})}
made from @var{ewoc}'s header and footer.
@end defun

@defun ewoc-set-hf ewoc header footer
This sets the header and footer of @var{ewoc} to the strings
@var{header} and @var{footer}, respectively.
@end defun

@defun ewoc-enter-first ewoc data
@defunx ewoc-enter-last ewoc data
These add a new node encapsulating @var{data}, putting it, respectively,
at the beginning or end of @var{ewoc}'s chain of nodes.
@end defun

@defun ewoc-enter-before ewoc node data
@defunx ewoc-enter-after ewoc node data
These add a new node encapsulating @var{data}, adding it to
@var{ewoc} before or after @var{node}, respectively.
@end defun

@defun ewoc-prev ewoc node
@defunx ewoc-next ewoc node
These return, respectively, the previous node and the next node of @var{node}
in @var{ewoc}.
@end defun

@defun ewoc-nth ewoc n
This returns the node in @var{ewoc} found at zero-based index @var{n}.
A negative @var{n} means count from the end.  @code{ewoc-nth} returns
@code{nil} if @var{n} is out of range.
@end defun

@defun ewoc-data node
This extracts the data encapsulated by @var{node} and returns it.
@end defun

@defun ewoc-set-data node data
This sets the data encapsulated by @var{node} to @var{data}.
@end defun

@defun ewoc-locate ewoc &optional pos guess
This determines the node in @var{ewoc} which contains point (or
@var{pos} if specified), and returns that node.  If @var{ewoc} has no
nodes, it returns @code{nil}.  If @var{pos} is before the first node,
it returns the first node; if @var{pos} is after the last node, it returns
the last node.  The optional third arg @var{guess}
should be a node that is likely to be near @var{pos}; this doesn't
alter the result, but makes the function run faster.
@end defun

@defun ewoc-location node
This returns the start position of @var{node}.
@end defun

@defun ewoc-goto-prev ewoc arg
@defunx ewoc-goto-next ewoc arg
These move point to the previous or next, respectively, @var{arg}th node
in @var{ewoc}.  @code{ewoc-goto-prev} does not move if it is already at
the first node or if @var{ewoc} is empty, whereas @code{ewoc-goto-next}
moves past the last node, returning @code{nil}.  Excepting this special
case, these functions return the node moved to.
@end defun

@defun ewoc-goto-node ewoc node
This moves point to the start of @var{node} in @var{ewoc}.
@end defun

@defun ewoc-refresh ewoc
This function regenerates the text of @var{ewoc}.  It works by
deleting the text between the header and the footer, i.e., all the
data elements' representations, and then calling the pretty-printer
function for each node, one by one, in order.
@end defun

@defun ewoc-invalidate ewoc &rest nodes
This is similar to @code{ewoc-refresh}, except that only @var{nodes} in
@var{ewoc} are updated instead of the entire set.
@end defun

@defun ewoc-delete ewoc &rest nodes
This deletes each node in @var{nodes} from @var{ewoc}.
@end defun

@defun ewoc-filter ewoc predicate &rest args
This calls @var{predicate} for each data element in @var{ewoc} and
deletes those nodes for which @var{predicate} returns @code{nil}.
Any @var{args} are passed to @var{predicate}.
@end defun

@defun ewoc-collect ewoc predicate &rest args
This calls @var{predicate} for each data element in @var{ewoc}
and returns a list of those elements for which @var{predicate}
returns non-@code{nil}.  The elements in the list are ordered
as in the buffer.  Any @var{args} are passed to @var{predicate}.
@end defun

@defun ewoc-map map-function ewoc &rest args
This calls @var{map-function} for each data element in @var{ewoc} and
updates those nodes for which @var{map-function} returns non-@code{nil}.
Any @var{args} are passed to @var{map-function}.
@end defun

@node Abstract Display Example
@subsection Abstract Display Example

  Here is a simple example using functions of the ewoc package to
implement a ``color components display'', an area in a buffer that
represents a vector of three integers (itself representing a 24-bit RGB
value) in various ways.

@example
(setq colorcomp-ewoc nil
      colorcomp-data nil
      colorcomp-mode-map nil
      colorcomp-labels ["Red" "Green" "Blue"])

(defun colorcomp-pp (data)
  (if data
      (let ((comp (aref colorcomp-data data)))
        (insert (aref colorcomp-labels data) "\t: #x"
                (format "%02X" comp) " "
                (make-string (ash comp -2) ?#) "\n"))
    (let ((cstr (format "#%02X%02X%02X"
                        (aref colorcomp-data 0)
                        (aref colorcomp-data 1)
                        (aref colorcomp-data 2)))
          (samp " (sample text) "))
      (insert "Color\t: "
              (propertize samp 'face
                          `(foreground-color . ,cstr))
              (propertize samp 'face
                          `(background-color . ,cstr))
              "\n"))))

(defun colorcomp (color)
  "Allow fiddling with COLOR in a new buffer.
The buffer is in Color Components mode."
  (interactive "sColor (name or #RGB or #RRGGBB): ")
  (when (string= "" color)
    (setq color "green"))
  (unless (color-values color)
    (error "No such color: %S" color))
  (switch-to-buffer
   (generate-new-buffer (format "originally: %s" color)))
  (kill-all-local-variables)
  (setq major-mode 'colorcomp-mode
        mode-name "Color Components")
  (use-local-map colorcomp-mode-map)
  (erase-buffer)
  (buffer-disable-undo)
  (let ((data (apply 'vector (mapcar (lambda (n) (ash n -8))
                                     (color-values color))))
        (ewoc (ewoc-create 'colorcomp-pp
                           "\nColor Components\n\n"
                           (substitute-command-keys
                            "\n\\@{colorcomp-mode-map@}"))))
    (set (make-local-variable 'colorcomp-data) data)
    (set (make-local-variable 'colorcomp-ewoc) ewoc)
    (ewoc-enter-last ewoc 0)
    (ewoc-enter-last ewoc 1)
    (ewoc-enter-last ewoc 2)
    (ewoc-enter-last ewoc nil)))
@end example

@cindex controller part, model/view/controller
  This example can be extended to be a ``color selection widget'' (in
other words, the controller part of the ``model/view/controller''
design paradigm) by defining commands to modify @code{colorcomp-data}
and to ``finish'' the selection process, and a keymap to tie it all
together conveniently.

@smallexample
(defun colorcomp-mod (index limit delta)
  (let ((cur (aref colorcomp-data index)))
    (unless (= limit cur)
      (aset colorcomp-data index (+ cur delta)))
    (ewoc-invalidate
     colorcomp-ewoc
     (ewoc-nth colorcomp-ewoc index)
     (ewoc-nth colorcomp-ewoc -1))))

(defun colorcomp-R-more () (interactive) (colorcomp-mod 0 255 1))
(defun colorcomp-G-more () (interactive) (colorcomp-mod 1 255 1))
(defun colorcomp-B-more () (interactive) (colorcomp-mod 2 255 1))
(defun colorcomp-R-less () (interactive) (colorcomp-mod 0 0 -1))
(defun colorcomp-G-less () (interactive) (colorcomp-mod 1 0 -1))
(defun colorcomp-B-less () (interactive) (colorcomp-mod 2 0 -1))

(defun colorcomp-copy-as-kill-and-exit ()
  "Copy the color components into the kill ring and kill the buffer.
The string is formatted #RRGGBB (hash followed by six hex digits)."
  (interactive)
  (kill-new (format "#%02X%02X%02X"
                    (aref colorcomp-data 0)
                    (aref colorcomp-data 1)
                    (aref colorcomp-data 2)))
  (kill-buffer nil))

(setq colorcomp-mode-map
      (let ((m (make-sparse-keymap)))
        (suppress-keymap m)
        (define-key m "i" 'colorcomp-R-less)
        (define-key m "o" 'colorcomp-R-more)
        (define-key m "k" 'colorcomp-G-less)
        (define-key m "l" 'colorcomp-G-more)
        (define-key m "," 'colorcomp-B-less)
        (define-key m "." 'colorcomp-B-more)
        (define-key m " " 'colorcomp-copy-as-kill-and-exit)
        m))
@end smallexample

Note that we never modify the data in each node, which is fixed when the
ewoc is created to be either @code{nil} or an index into the vector
@code{colorcomp-data}, the actual color components.

@node Blinking
@section Blinking Parentheses
@cindex parenthesis matching
@cindex blinking parentheses
@cindex balancing parentheses

  This section describes the mechanism by which Emacs shows a matching
open parenthesis when the user inserts a close parenthesis.

@defvar blink-paren-function
The value of this variable should be a function (of no arguments) to
be called whenever a character with close parenthesis syntax is inserted.
The value of @code{blink-paren-function} may be @code{nil}, in which
case nothing is done.
@end defvar

@defopt blink-matching-paren
If this variable is @code{nil}, then @code{blink-matching-open} does
nothing.
@end defopt

@defopt blink-matching-paren-distance
This variable specifies the maximum distance to scan for a matching
parenthesis before giving up.
@end defopt

@defopt blink-matching-delay
This variable specifies the number of seconds for the cursor to remain
at the matching parenthesis.  A fraction of a second often gives
good results, but the default is 1, which works on all systems.
@end defopt

@deffn Command blink-matching-open
This function is the default value of @code{blink-paren-function}.  It
assumes that point follows a character with close parenthesis syntax and
moves the cursor momentarily to the matching opening character.  If that
character is not already on the screen, it displays the character's
context in the echo area.  To avoid long delays, this function does not
search farther than @code{blink-matching-paren-distance} characters.

Here is an example of calling this function explicitly.

@smallexample
@group
(defun interactive-blink-matching-open ()
  "Indicate momentarily the start of parenthesized sexp before point."
  (interactive)
@end group
@group
  (let ((blink-matching-paren-distance
         (buffer-size))
        (blink-matching-paren t))
    (blink-matching-open)))
@end group
@end smallexample
@end deffn

@node Character Display
@section Character Display

  This section describes how characters are actually displayed by
Emacs.  Typically, a character is displayed as a @dfn{glyph} (a
graphical symbol which occupies one character position on the screen),
whose appearance corresponds to the character itself.  For example,
the character @samp{a} (character code 97) is displayed as @samp{a}.
Some characters, however, are displayed specially.  For example, the
formfeed character (character code 12) is usually displayed as a
sequence of two glyphs, @samp{^L}, while the newline character
(character code 10) starts a new screen line.

  You can modify how each character is displayed by defining a
@dfn{display table}, which maps each character code into a sequence of
glyphs.  @xref{Display Tables}.

@menu
* Usual Display::       The usual conventions for displaying characters.
* Display Tables::      What a display table consists of.
* Active Display Table::  How Emacs selects a display table to use.
* Glyphs::              How to define a glyph, and what glyphs mean.
* Glyphless Chars::     How glyphless characters are drawn.
@end menu

@node Usual Display
@subsection Usual Display Conventions

  Here are the conventions for displaying each character code (in the
absence of a display table, which can override these
@iftex
conventions).
@end iftex
@ifnottex
conventions; @pxref{Display Tables}).
@end ifnottex

@cindex printable ASCII characters
@itemize @bullet
@item
The @dfn{printable @acronym{ASCII} characters}, character codes 32
through 126 (consisting of numerals, English letters, and symbols like
@samp{#}) are displayed literally.

@item
The tab character (character code 9) displays as whitespace stretching
up to the next tab stop column.  @xref{Text Display,,, emacs, The GNU
Emacs Manual}.  The variable @code{tab-width} controls the number of
spaces per tab stop (see below).

@item
The newline character (character code 10) has a special effect: it
ends the preceding line and starts a new line.

@cindex ASCII control characters
@item
The non-printable @dfn{@acronym{ASCII} control characters}---character
codes 0 through 31, as well as the @key{DEL} character (character code
127)---display in one of two ways according to the variable
@code{ctl-arrow}.  If this variable is non-@code{nil} (the default),
these characters are displayed as sequences of two glyphs, where the
first glyph is @samp{^} (a display table can specify a glyph to use
instead of @samp{^}); e.g., the @key{DEL} character is displayed as
@samp{^?}.

If @code{ctl-arrow} is @code{nil}, these characters are displayed as
octal escapes (see below).

This rule also applies to carriage return (character code 13), if that
character appears in the buffer.  But carriage returns usually do not
appear in buffer text; they are eliminated as part of end-of-line
conversion (@pxref{Coding System Basics}).

@cindex octal escapes
@item
@dfn{Raw bytes} are non-@acronym{ASCII} characters with codes 128
through 255 (@pxref{Text Representations}).  These characters display
as @dfn{octal escapes}: sequences of four glyphs, where the first
glyph is the @acronym{ASCII} code for @samp{\}, and the others are
digit characters representing the character code in octal.  (A display
table can specify a glyph to use instead of @samp{\}.)

@item
Each non-@acronym{ASCII} character with code above 255 is displayed
literally, if the terminal supports it.  If the terminal does not
support it, the character is said to be @dfn{glyphless}, and it is
usually displayed using a placeholder glyph.  For example, if a
graphical terminal has no font for a character, Emacs usually displays
a box containing the character code in hexadecimal.  @xref{Glyphless
Chars}.
@end itemize

  The above display conventions apply even when there is a display
table, for any character whose entry in the active display table is
@code{nil}.  Thus, when you set up a display table, you need only
specify the characters for which you want special behavior.

  The following variables affect how certain characters are displayed
on the screen.  Since they change the number of columns the characters
occupy, they also affect the indentation functions.  They also affect
how the mode line is displayed; if you want to force redisplay of the
mode line using the new values, call the function
@code{force-mode-line-update} (@pxref{Mode Line Format}).

@defopt ctl-arrow
@cindex control characters in display
This buffer-local variable controls how control characters are
displayed.  If it is non-@code{nil}, they are displayed as a caret
followed by the character: @samp{^A}.  If it is @code{nil}, they are
displayed as octal escapes: a backslash followed by three octal
digits, as in @samp{\001}.
@end defopt

@defopt tab-width
The value of this buffer-local variable is the spacing between tab
stops used for displaying tab characters in Emacs buffers.  The value
is in units of columns, and the default is 8.  Note that this feature
is completely independent of the user-settable tab stops used by the
command @code{tab-to-tab-stop}.  @xref{Indent Tabs}.
@end defopt

@node Display Tables
@subsection Display Tables

@cindex display table
  A display table is a special-purpose char-table
(@pxref{Char-Tables}), with @code{display-table} as its subtype, which
is used to override the usual character display conventions.  This
section describes how to make, inspect, and assign elements to a
display table object.

@defun make-display-table
This creates and returns a display table.  The table initially has
@code{nil} in all elements.
@end defun

  The ordinary elements of the display table are indexed by character
codes; the element at index @var{c} says how to display the character
code @var{c}.  The value should be @code{nil} (which means to display
the character @var{c} according to the usual display conventions;
@pxref{Usual Display}), or a vector of glyph codes (which means to
display the character @var{c} as those glyphs; @pxref{Glyphs}).

  @strong{Warning:} if you use the display table to change the display
of newline characters, the whole buffer will be displayed as one long
``line''.

  The display table also has six ``extra slots'' which serve special
purposes.  Here is a table of their meanings; @code{nil} in any slot
means to use the default for that slot, as stated below.

@table @asis
@item 0
The glyph for the end of a truncated screen line (the default for this
is @samp{$}).  @xref{Glyphs}.  On graphical terminals, Emacs uses
arrows in the fringes to indicate truncation, so the display table has
no effect.

@item 1
The glyph for the end of a continued line (the default is @samp{\}).
On graphical terminals, Emacs uses curved arrows in the fringes to
indicate continuation, so the display table has no effect.

@item 2
The glyph for indicating a character displayed as an octal character
code (the default is @samp{\}).

@item 3
The glyph for indicating a control character (the default is @samp{^}).

@item 4
A vector of glyphs for indicating the presence of invisible lines (the
default is @samp{...}).  @xref{Selective Display}.

@item 5
The glyph used to draw the border between side-by-side windows (the
default is @samp{|}).  @xref{Splitting Windows}.  This takes effect only
when there are no scroll bars; if scroll bars are supported and in use,
a scroll bar separates the two windows.
@end table

  For example, here is how to construct a display table that mimics
the effect of setting @code{ctl-arrow} to a non-@code{nil} value
(@pxref{Glyphs}, for the function @code{make-glyph-code}):

@example
(setq disptab (make-display-table))
(dotimes (i 32)
  (or (= i ?\t)
      (= i ?\n)
      (aset disptab i
            (vector (make-glyph-code ?^ 'escape-glyph)
                    (make-glyph-code (+ i 64) 'escape-glyph)))))
(aset disptab 127
      (vector (make-glyph-code ?^ 'escape-glyph)
              (make-glyph-code ?? 'escape-glyph)))))
@end example

@defun display-table-slot display-table slot
This function returns the value of the extra slot @var{slot} of
@var{display-table}.  The argument @var{slot} may be a number from 0 to
5 inclusive, or a slot name (symbol).  Valid symbols are
@code{truncation}, @code{wrap}, @code{escape}, @code{control},
@code{selective-display}, and @code{vertical-border}.
@end defun

@defun set-display-table-slot display-table slot value
This function stores @var{value} in the extra slot @var{slot} of
@var{display-table}.  The argument @var{slot} may be a number from 0 to
5 inclusive, or a slot name (symbol).  Valid symbols are
@code{truncation}, @code{wrap}, @code{escape}, @code{control},
@code{selective-display}, and @code{vertical-border}.
@end defun

@defun describe-display-table display-table
This function displays a description of the display table
@var{display-table} in a help buffer.
@end defun

@deffn Command describe-current-display-table
This command displays a description of the current display table in a
help buffer.
@end deffn

@node Active Display Table
@subsection Active Display Table
@cindex active display table

  Each window can specify a display table, and so can each buffer.
The window's display table, if there is one, takes precedence over the
buffer's display table.  If neither exists, Emacs tries to use the
standard display table; if that is @code{nil}, Emacs uses the usual
character display conventions (@pxref{Usual Display}).

  Note that display tables affect how the mode line is displayed, so
if you want to force redisplay of the mode line using a new display
table, call @code{force-mode-line-update} (@pxref{Mode Line Format}).

@defun window-display-table &optional window
This function returns @var{window}'s display table, or @code{nil} if
there is none.  The default for @var{window} is the selected window.
@end defun

@defun set-window-display-table window table
This function sets the display table of @var{window} to @var{table}.
The argument @var{table} should be either a display table or
@code{nil}.
@end defun

@defvar buffer-display-table
This variable is automatically buffer-local in all buffers; its value
specifies the buffer's display table.  If it is @code{nil}, there is
no buffer display table.
@end defvar

@defvar standard-display-table
The value of this variable is the standard display table, which is
used when Emacs is displaying a buffer in a window with neither a
window display table nor a buffer display table defined.  Its default
is @code{nil}.
@end defvar

The @file{disp-table} library defines several functions for changing
the standard display table.

@node Glyphs
@subsection Glyphs
@cindex glyph

@cindex glyph code
  A @dfn{glyph} is a graphical symbol which occupies a single
character position on the screen.  Each glyph is represented in Lisp
as a @dfn{glyph code}, which specifies a character and optionally a
face to display it in (@pxref{Faces}).  The main use of glyph codes is
as the entries of display tables (@pxref{Display Tables}).  The
following functions are used to manipulate glyph codes:

@defun make-glyph-code char &optional face
This function returns a glyph code representing char @var{char} with
face @var{face}.  If @var{face} is omitted or @code{nil}, the glyph
uses the default face; in that case, the glyph code is an integer.  If
@var{face} is non-@code{nil}, the glyph code is not necessarily an
integer object.
@end defun

@defun glyph-char glyph
This function returns the character of glyph code @var{glyph}.
@end defun

@defun glyph-face glyph
This function returns face of glyph code @var{glyph}, or @code{nil} if
@var{glyph} uses the default face.
@end defun

@ifnottex
  You can set up a @dfn{glyph table} to change how glyph codes are
actually displayed on text terminals.  This feature is semi-obsolete;
use @code{glyphless-char-display} instead (@pxref{Glyphless Chars}).

@defvar glyph-table
The value of this variable, if non-@code{nil}, is the current glyph
table.  It takes effect only on character terminals; on graphical
displays, all glyphs are displayed literally.  The glyph table should
be a vector whose @var{g}th element specifies how to display glyph
code @var{g}, where @var{g} is the glyph code for a glyph whose face
is unspecified.  Each element should be one of the following:

@table @asis
@item @code{nil}
Display this glyph literally.

@item a string
Display this glyph by sending the specified string to the terminal.

@item a glyph code
Display the specified glyph code instead.
@end table

Any integer glyph code greater than or equal to the length of the
glyph table is displayed literally.
@end defvar
@end ifnottex

@node Glyphless Chars
@subsection Glyphless Character Display
@cindex glyphless characters

  @dfn{Glyphless characters} are characters which are displayed in a
special way, e.g., as a box containing a hexadecimal code, instead of
being displayed literally.  These include characters which are
explicitly defined to be glyphless, as well as characters for which
there is no available font (on a graphical display), and characters
which cannot be encoded by the terminal's coding system (on a text
terminal).

@defvar glyphless-char-display
The value of this variable is a char-table which defines glyphless
characters and how they are displayed.  Each entry must be one of the
following display methods:

@table @asis
@item @code{nil}
Display the character in the usual way.

@item @code{zero-width}
Don't display the character.

@item @code{thin-space}
Display a thin space, 1-pixel wide on graphical displays, or
1-character wide on text terminals.

@item @code{empty-box}
Display an empty box.

@item @code{hex-code}
Display a box containing the Unicode codepoint of the character, in
hexadecimal notation.

@item an @acronym{ASCII} string
Display a box containing that string.

@item a cons cell @code{(@var{graphical} . @var{text})}
Display with @var{graphical} on graphical displays, and with
@var{text} on text terminals.  Both @var{graphical} and @var{text}
must be one of the display methods described above.
@end table

@noindent
The @code{thin-space}, @code{empty-box}, @code{hex-code}, and
@acronym{ASCII} string display methods are drawn with the
@code{glyphless-char} face.

The char-table has one extra slot, which determines how to display any
character that cannot be displayed with any available font, or cannot
be encoded by the terminal's coding system.  Its value should be one
of the above display methods, except @code{zero-width} or a cons cell.

If a character has a non-@code{nil} entry in an active display table,
the display table takes effect; in this case, Emacs does not consult
@code{glyphless-char-display} at all.
@end defvar

@defopt glyphless-char-display-control
This user option provides a convenient way to set
@code{glyphless-char-display} for groups of similar characters.  Do
not set its value directly from Lisp code; the value takes effect only
via a custom @code{:set} function (@pxref{Variable Definitions}),
which updates @code{glyphless-char-display}.

Its value should be an alist of elements @code{(@var{group}
. @var{method})}, where @var{group} is a symbol specifying a group of
characters, and @var{method} is a symbol specifying how to display
them.

@var{group} should be one of the following:

@table @code
@item c0-control
@acronym{ASCII} control characters @code{U+0000} to @code{U+001F},
excluding the newline and tab characters (normally displayed as escape
sequences like @samp{^A}; @pxref{Text Display,, How Text Is Displayed,
emacs, The GNU Emacs Manual}).

@item c1-control
Non-@acronym{ASCII}, non-printing characters @code{U+0080} to
@code{U+009F} (normally displayed as octal escape sequences like
@samp{\230}).

@item format-control
Characters of Unicode General Category `Cf', such as @samp{U+200E}
(Left-to-Right Mark), but excluding characters that have graphic
images, such as @samp{U+00AD} (Soft Hyphen).

@item no-font
Characters for there is no suitable font, or which cannot be encoded
by the terminal's coding system.
@end table

@c FIXME: this can also be `acronym', but that's not currently
@c completely implemented; it applies only to the format-control
@c group, and only works if the acronym is in `char-acronym-table'.
The @var{method} symbol should be one of @code{zero-width},
@code{thin-space}, @code{empty-box}, or @code{hex-code}.  These have
the same meanings as in @code{glyphless-char-display}, above.
@end defopt

@node Beeping
@section Beeping
@cindex bell

  This section describes how to make Emacs ring the bell (or blink the
screen) to attract the user's attention.  Be conservative about how
often you do this; frequent bells can become irritating.  Also be
careful not to use just beeping when signaling an error is more
appropriate (@pxref{Errors}).

@defun ding &optional do-not-terminate
@cindex keyboard macro termination
This function beeps, or flashes the screen (see @code{visible-bell} below).
It also terminates any keyboard macro currently executing unless
@var{do-not-terminate} is non-@code{nil}.
@end defun

@defun beep &optional do-not-terminate
This is a synonym for @code{ding}.
@end defun

@defopt visible-bell
This variable determines whether Emacs should flash the screen to
represent a bell.  Non-@code{nil} means yes, @code{nil} means no.
This is effective on graphical displays, and on text terminals
provided the terminal's Termcap entry defines the visible bell
capability (@samp{vb}).
@end defopt

@defvar ring-bell-function
If this is non-@code{nil}, it specifies how Emacs should ``ring the
bell''.  Its value should be a function of no arguments.  If this is
non-@code{nil}, it takes precedence over the @code{visible-bell}
variable.
@end defvar

@node Window Systems
@section Window Systems

  Emacs works with several window systems, most notably the X Window
System.  Both Emacs and X use the term ``window'', but use it
differently.  An Emacs frame is a single window as far as X is
concerned; the individual Emacs windows are not known to X at all.

@defvar window-system
This terminal-local variable tells Lisp programs what window system
Emacs is using for displaying the frame.  The possible values are

@table @code
@item x
@cindex X Window System
Emacs is displaying the frame using X.
@item w32
Emacs is displaying the frame using native MS-Windows GUI.
@item ns
Emacs is displaying the frame using the Nextstep interface (used on
GNUstep and Mac OS X).
@item pc
Emacs is displaying the frame using MS-DOS direct screen writes.
@item nil
Emacs is displaying the frame on a character-based terminal.
@end table
@end defvar

@defvar initial-window-system
This variable holds the value of @code{window-system} used for the
first frame created by Emacs during startup.  (When Emacs is invoked
with the @option{--daemon} option, it does not create any initial
frames, so @code{initial-window-system} is @code{nil}.  @xref{Initial
Options, daemon,, emacs, The GNU Emacs Manual}.)
@end defvar

@defun window-system &optional frame
This function returns a symbol whose name tells what window system is
used for displaying @var{frame} (which defaults to the currently
selected frame).  The list of possible symbols it returns is the same
one documented for the variable @code{window-system} above.
@end defun

  Do @emph{not} use @code{window-system} and
@code{initial-window-system} as predicates or boolean flag variables,
if you want to write code that works differently on text terminals and
graphic displays.  That is because @code{window-system} is not a good
indicator of Emacs capabilities on a given display type.  Instead, use
@code{display-graphic-p} or any of the other @code{display-*-p}
predicates described in @ref{Display Feature Testing}.

@defvar window-setup-hook
This variable is a normal hook which Emacs runs after handling the
initialization files.  Emacs runs this hook after it has completed
loading your init file, the default initialization file (if
any), and the terminal-specific Lisp code, and running the hook
@code{emacs-startup-hook}.

This hook is used for internal purposes: setting up communication with
the window system, and creating the initial window.  Users should not
interfere with it.
@end defvar

@node Bidirectional Display
@section Bidirectional Display
@cindex bidirectional display
@cindex right-to-left text

  Emacs can display text written in scripts, such as Arabic, Farsi,
and Hebrew, whose natural ordering for horizontal text display runs
from right to left.  Furthermore, segments of Latin script and digits
embedded in right-to-left text are displayed left-to-right, while
segments of right-to-left script embedded in left-to-right text
(e.g., Arabic or Hebrew text in comments or strings in a program
source file) are appropriately displayed right-to-left.  We call such
mixtures of left-to-right and right-to-left text @dfn{bidirectional
text}.  This section describes the facilities and options for editing
and displaying bidirectional text.

@cindex logical order
@cindex reading order
@cindex visual order
@cindex unicode bidirectional algorithm
@cindex UBA
@cindex bidirectional reordering
@cindex reordering, of bidirectional text
  Text is stored in Emacs buffers and strings in @dfn{logical} (or
@dfn{reading}) order, i.e., the order in which a human would read
each character.  In right-to-left and bidirectional text, the order in
which characters are displayed on the screen (called @dfn{visual
order}) is not the same as logical order; the characters' screen
positions do not increase monotonically with string or buffer
position.  In performing this @dfn{bidirectional reordering}, Emacs
follows the Unicode Bidirectional Algorithm (a.k.a.@: @acronym{UBA}),
which is described in Annex #9 of the Unicode standard
(@url{http://www.unicode.org/reports/tr9/}).  Emacs provides a ``Full
Bidirectionality'' class implementation of the @acronym{UBA}.

@defvar bidi-display-reordering
If the value of this buffer-local variable is non-@code{nil} (the
default), Emacs performs bidirectional reordering for display.  The
reordering affects buffer text, as well as display strings and overlay
strings from text and overlay properties in the buffer (@pxref{Overlay
Properties}, and @pxref{Display Property}).  If the value is
@code{nil}, Emacs does not perform bidirectional reordering in the
buffer.

The default value of @code{bidi-display-reordering} controls the
reordering of strings which are not directly supplied by a buffer,
including the text displayed in mode lines (@pxref{Mode Line Format})
and header lines (@pxref{Header Lines}).
@end defvar

@cindex unibyte buffers, and bidi reordering
  Emacs never reorders the text of a unibyte buffer, even if
@code{bidi-display-reordering} is non-@code{nil} in the buffer.  This
is because unibyte buffers contain raw bytes, not characters, and thus
lack the directionality properties required for reordering.
Therefore, to test whether text in a buffer will be reordered for
display, it is not enough to test the value of
@code{bidi-display-reordering} alone.  The correct test is this:

@example
 (if (and enable-multibyte-characters
          bidi-display-reordering)
     ;; Buffer is being reordered for display
   )
@end example

  However, unibyte display and overlay strings @emph{are} reordered if
their parent buffer is reordered.  This is because plain-@sc{ascii}
strings are stored by Emacs as unibyte strings.  If a unibyte display
or overlay string includes non-@sc{ascii} characters, these characters
are assumed to have left-to-right direction.

@cindex display properties, and bidi reordering of text
  Text covered by @code{display} text properties, by overlays with
@code{display} properties whose value is a string, and by any other
properties that replace buffer text, is treated as a single unit when
it is reordered for display.  That is, the entire chunk of text
covered by these properties is reordered together.  Moreover, the
bidirectional properties of the characters in such a chunk of text are
ignored, and Emacs reorders them as if they were replaced with a
single character @code{U+FFFC}, known as the @dfn{Object Replacement
Character}.  This means that placing a display property over a portion
of text may change the way that the surrounding text is reordered for
display.  To prevent this unexpected effect, always place such
properties on text whose directionality is identical with text that
surrounds it.

@cindex base direction of a paragraph
  Each paragraph of bidirectional text has a @dfn{base direction},
either right-to-left or left-to-right.  Left-to-right paragraphs are
displayed beginning at the left margin of the window, and are
truncated or continued when the text reaches the right margin.
Right-to-left paragraphs are displayed beginning at the right margin,
and are continued or truncated at the left margin.

  By default, Emacs determines the base direction of each paragraph by
looking at the text at its beginning.  The precise method of
determining the base direction is specified by the @acronym{UBA}; in a
nutshell, the first character in a paragraph that has an explicit
directionality determines the base direction of the paragraph.
However, sometimes a buffer may need to force a certain base direction
for its paragraphs.  For example, buffers containing program source
code should force all paragraphs to be displayed left-to-right.  You
can use following variable to do this:

@defvar bidi-paragraph-direction
If the value of this buffer-local variable is the symbol
@code{right-to-left} or @code{left-to-right}, all paragraphs in the
buffer are assumed to have that specified direction.  Any other value
is equivalent to @code{nil} (the default), which means to determine
the base direction of each paragraph from its contents.

@cindex @code{prog-mode}, and @code{bidi-paragraph-direction}
Modes for program source code should set this to @code{left-to-right}.
Prog mode does this by default, so modes derived from Prog mode do not
need to set this explicitly (@pxref{Basic Major Modes}).
@end defvar

@defun current-bidi-paragraph-direction &optional buffer
This function returns the paragraph direction at point in the named
@var{buffer}.  The returned value is a symbol, either
@code{left-to-right} or @code{right-to-left}.  If @var{buffer} is
omitted or @code{nil}, it defaults to the current buffer.  If the
buffer-local value of the variable @code{bidi-paragraph-direction} is
non-@code{nil}, the returned value will be identical to that value;
otherwise, the returned value reflects the paragraph direction
determined dynamically by Emacs.  For buffers whose value of
@code{bidi-display-reordering} is @code{nil} as well as unibyte
buffers, this function always returns @code{left-to-right}.
@end defun

@cindex visual-order cursor motion
  Sometimes there's a need to move point in strict visual order,
either to the left or to the right of its current screen position.
Emacs provides a primitive to do that.

@defun move-point-visually direction
This function moves point of the currently selected window to the
buffer position that appears immediately to the right or to the left
of point on the screen.  If @var{direction} is positive, point will
move one screen position to the right, otherwise it will move one
screen position to the left.  Note that, depending on the surrounding
bidirectional context, this could potentially move point many buffer
positions away.  If invoked at the end of a screen line, the function
moves point to the rightmost or leftmost screen position of the next
or previous screen line, as appropriate for the value of
@var{direction}.

The function returns the new buffer position as its value.
@end defun

@cindex layout on display, and bidirectional text
@cindex jumbled display of bidirectional text
@cindex concatenating bidirectional strings
  Bidirectional reordering can have surprising and unpleasant effects
when two strings with bidirectional content are juxtaposed in a
buffer, or otherwise programmatically concatenated into a string of
text.  A typical problematic case is when a buffer consists of
sequences of text ``fields'' separated by whitespace or punctuation
characters, like Buffer Menu mode or Rmail Summary Mode.  Because the
punctuation characters used as separators have @dfn{weak
directionality}, they take on the directionality of surrounding text.
As result, a numeric field that follows a field with bidirectional
content can be displayed @emph{to the left} of the preceding field,
messing up the expected layout.  There are several ways to avoid this
problem:

@itemize @minus
@item
Append the special character @code{U+200E}, LEFT-TO-RIGHT MARK, or
@acronym{LRM}, to the end of each field that may have bidirectional
content, or prepend it to the beginning of the following field.  The
function @code{bidi-string-mark-left-to-right}, described below, comes
in handy for this purpose.  (In a right-to-left paragraph, use
@code{U+200F}, RIGHT-TO-LEFT MARK, or @acronym{RLM}, instead.)  This
is one of the solutions recommended by the UBA.

@item
Include the tab character in the field separator.  The tab character
plays the role of @dfn{segment separator} in bidirectional reordering,
causing the text on either side to be reordered separately.

@cindex @code{space} display spec, and bidirectional text
@item
Separate fields with a @code{display} property or overlay with a
property value of the form @code{(space . PROPS)} (@pxref{Specified
Space}).  Emacs treats this display specification as a @dfn{paragraph
separator}, and reorders the text on either side separately.
@end itemize

@defun bidi-string-mark-left-to-right string
This function returns its argument @var{string}, possibly modified,
such that the result can be safely concatenated with another string,
or juxtaposed with another string in a buffer, without disrupting the
relative layout of this string and the next one on display.  If the
string returned by this function is displayed as part of a
left-to-right paragraph, it will always appear on display to the left
of the text that follows it.  The function works by examining the
characters of its argument, and if any of those characters could cause
reordering on display, the function appends the @acronym{LRM}
character to the string.  The appended @acronym{LRM} character is made
invisible by giving it an @code{invisible} text property of @code{t}
(@pxref{Invisible Text}).
@end defun

  The reordering algorithm uses the bidirectional properties of the
characters stored as their @code{bidi-class} property
(@pxref{Character Properties}).  Lisp programs can change these
properties by calling the @code{put-char-code-property} function.
However, doing this requires a thorough understanding of the
@acronym{UBA}, and is therefore not recommended.  Any changes to the
bidirectional properties of a character have global effect: they
affect all Emacs frames and windows.

  Similarly, the @code{mirroring} property is used to display the
appropriate mirrored character in the reordered text.  Lisp programs
can affect the mirrored display by changing this property.  Again, any
such changes affect all of Emacs display.