The text you are editing in Emacs resides in an object called a buffer. Each time you visit a file, a buffer is used to hold the file’s text. Each time you invoke Dired, a buffer is used to hold the directory listing. If you send a message with C-x m, a buffer is used to hold the text of the message. When you ask for a command’s documentation, that appears in a buffer named *Help*.
Buffers exist as long as they are in use, and are deleted (“killed”) when no longer needed, either by you (see Killing Buffers) or by Emacs (e.g., when you exit Emacs, see Exiting Emacs).
Each buffer has a unique name, which can be of any length. When a buffer is displayed in a window, its name is shown in the mode line (see The Mode Line). The distinction between upper and lower case matters in buffer names. Most buffers are made by visiting files, and their names are derived from the files’ names; however, you can also create an empty buffer with any name you want. A newly started Emacs has several buffers, including one named *scratch*, which can be used for evaluating Lisp expressions and is not associated with any file (see Lisp Interaction Buffers).
At any time, one and only one buffer is selected; we call it the current buffer. We sometimes say that a command operates on “the buffer”; this really means that it operates on the current buffer. When there is only one Emacs window, the buffer displayed in that window is current. When there are multiple windows, the buffer displayed in the selected window is current. See Multiple Windows.
A buffer’s contents consist of a series of characters, each of which optionally carries a set of text properties (see Text properties) that can specify more information about that character.
Aside from its textual contents, each buffer records several pieces of information, such as what file it is visiting (if any), whether it is modified, and what major mode and minor modes are in effect (see Major and Minor Modes). These are stored in buffer-local variables—variables that can have a different value in each buffer. See Local Variables.
A buffer’s size cannot be larger than some maximum, which is defined by the largest buffer position representable by Emacs integers. This is because Emacs tracks buffer positions using that data type. For typical 64-bit machines, this maximum buffer size is 2^{61} - 2 bytes, or about 2 EiB. For typical 32-bit machines, the maximum is usually 2^{29} - 2 bytes, or about 512 MiB. Buffer sizes are also limited by the amount of memory in the system.
Select or create a buffer named buffer (switch-to-buffer).
Similar, but select buffer in another window
(switch-to-buffer-other-window).
Similar, but select buffer in a separate frame
(switch-to-buffer-other-frame).
Select the previous buffer in the buffer list (previous-buffer).
Select the next buffer in the buffer list (next-buffer).
Read a number n and move to line n in the most recently selected buffer other than the current buffer, in another window.
The C-x b (switch-to-buffer) command reads a buffer
name using the minibuffer. Then it makes that buffer current, and
displays it in the currently-selected window. An empty input
specifies the buffer that was current most recently among those not
now displayed in any window.
While entering the buffer name, you can use the usual completion and history commands (see The Minibuffer). Note that C-x b, and related commands, use permissive completion with confirmation for minibuffer completion: if you type RET when the minibuffer text names a nonexistent buffer, Emacs prints ‘[Confirm]’ and you must type a second RET to submit that buffer name. See Completion Exit, for details. For other completion options and features, see Completion Options.
If you specify a buffer that does not exist, C-x b creates a
new, empty buffer that is not visiting any file, and selects it for
editing. The default value of the variable major-mode
determines the new buffer’s major mode; the default value is
Fundamental mode. See Major Modes. One reason to create a new
buffer is to use it for making temporary notes. If you try to save
it, Emacs asks for the file name to use, and the buffer’s major mode
is re-established taking that file name into account (see Choosing File Modes).
For conveniently switching between a few buffers, use the commands
C-x LEFT and C-x RIGHT. C-x LEFT
(previous-buffer) selects the previous buffer (following the
order of most recent selection in the current frame), while C-x
RIGHT (next-buffer) moves through buffers in the reverse
direction. Both commands support a numeric prefix argument that
serves as a repeat count.
To select a buffer in a window other than the current one
(see Multiple Windows), type C-x 4 b
(switch-to-buffer-other-window). This prompts for a buffer
name using the minibuffer, displays that buffer in another window, and
selects that window.
Similarly, C-x 5 b (switch-to-buffer-other-frame)
prompts for a buffer name, displays that buffer in another frame
(see Frames and Graphical Displays), and selects that frame. If the buffer is already
being shown in a window on another frame, Emacs selects that window
and frame instead of creating a new frame.
See Displaying a Buffer in a Window, for how the C-x 4 b and C-x 5 b commands get the window and/or frame to display in.
In addition, C-x C-f, and any other command for visiting a file, can also be used to switch to an existing file-visiting buffer. See Visiting Files.
C-u M-g M-g, that is goto-line with a plain prefix
argument, reads a number n using the minibuffer, selects the
most recently selected buffer other than the current buffer in another
window, and then moves point to the beginning of line number n
in that buffer. This is mainly useful in a buffer that refers to line
numbers in another buffer: if point is on or just after a number,
goto-line uses that number as the default for n. Note
that prefix arguments other than just C-u behave differently.
C-u 4 M-g M-g goes to line 4 in the current buffer,
without reading a number from the minibuffer. (Remember that M-g
M-g without prefix argument reads a number n and then moves to
line number n in the current buffer. See Changing the Location of Point.)
Emacs uses buffer names that start with a space for internal purposes. It treats these buffers specially in minor ways—for example, by default they do not record undo information. It is best to avoid using such buffer names yourself.
List the existing buffers (list-buffers).
To display a list of existing buffers, type C-x C-b. This pops up a buffer menu in a buffer named *Buffer List*. Each line in the list shows one buffer’s name, size, major mode and visited file. The buffers are listed in the order that they were current; the buffers that were current most recently come first. This section describes how the list of buffers is displayed and how to interpret the various indications in the list; see Operating on Several Buffers, for description of the special mode in the *Buffer List* buffer and the commands available there.
‘.’ in the first field of a line indicates that the buffer is current. ‘%’ indicates a read-only buffer. ‘*’ indicates that the buffer is modified. If several buffers are modified, it may be time to save some with C-x s (see Commands for Saving Files). Here is an example of a buffer list:
CRM Buffer Size Mode File
. * .emacs 3294 Elisp/l ~/.emacs
% *Help* 101 Help
search.c 86055 C ~/cvs/emacs/src/search.c
% src 20959 Dired by name ~/cvs/emacs/src/
* *mail* 42 Mail
% HELLO 1607 Fundamental ~/cvs/emacs/etc/HELLO
% NEWS 481184 Outline ~/cvs/emacs/etc/NEWS
*scratch* 191 Lisp Interaction
* *Messages* 1554 Messages
The buffer *Help* was made by a help request (see Help); it
is not visiting any file. The buffer src was made by Dired on
the directory ~/cvs/emacs/src/. You can list only buffers that
are visiting files by giving the command a prefix argument, as in
C-u C-x C-b.
By default, list-buffers omits buffers whose names begin with a
space, unless they visit files: such buffers are used internally by
Emacs (but the I command countermands that, see Operating on Several Buffers).
Toggle read-only status of buffer (read-only-mode).
Change the name of the current buffer.
Rename the current buffer by adding ‘<number>’ to the end.
Scroll through buffer buffer. See View Mode.
A buffer can be read-only, which means that commands to insert or delete its text are not allowed. (However, other commands, like C-x RET f, can still mark it as modified, see Specifying a Coding System for File Text). The mode line indicates read-only buffers with ‘%%’ or ‘%*’ near the left margin. See The Mode Line. Read-only buffers are usually made by subsystems such as Dired and Rmail that have special commands to operate on the text. Visiting a file whose access control says you cannot write it also makes the buffer read-only.
The command C-x C-q (read-only-mode) makes a read-only
buffer writable, and makes a writable buffer read-only. This works by
setting the variable buffer-read-only, which has a local value
in each buffer and makes the buffer read-only if its value is
non-nil. If you change the option view-read-only to a
non-nil value, making the buffer read-only with C-x C-q
also enables View mode in the buffer (see View Mode).
C-x x r (rename-buffer changes the name of the current
buffer. You specify the new name as a minibuffer argument; there is
no default. If you specify a name that is in use for some other
buffer, an error happens and no renaming is done.
C-x x u (rename-uniquely) renames the current buffer to
a similar name with a numeric suffix added to make it both different
and unique. This command does not need an argument. It is useful for
creating multiple shell buffers: if you rename the *shell*
buffer, then do M-x shell again, it makes a new shell buffer
named *shell*; meanwhile, the old shell buffer continues to
exist under its new name. This method is also good for mail buffers,
compilation buffers, and most Emacs features that create special
buffers with particular names. (With some of these features, such as
M-x compile, M-x grep, you need to switch to some other
buffer before using the command again, otherwise it will reuse the
current buffer despite the name change.)
The commands M-x append-to-buffer and C-x x i
(insert-buffer) can also be used to copy text from one buffer
to another. See Accumulating Text.
If you continue an Emacs session for a while, you may accumulate a large number of buffers. You may then find it convenient to kill the buffers you no longer need. (Some other editors call this operation close, and talk about “closing the buffer” or “closing the file” visited in the buffer.) On most operating systems, killing a buffer releases the memory Emacs used for the buffer back to the operating system so that other programs can use it. Here are some commands for killing buffers:
Kill buffer buffer (kill-buffer).
Offer to kill each buffer, one by one.
Offer to kill all buffers matching a regular expression.
Like kill-matching-buffers, but don’t ask for confirmation.
C-x k (kill-buffer) kills one buffer, whose name you
specify in the minibuffer. The default, used if you type just
RET in the minibuffer, is to kill the current buffer. If you
kill the current buffer, another buffer becomes current: one that was
current in the recent past but is not displayed in any window now. If
you ask to kill a file-visiting buffer that is modified, then you must
confirm with yes before the buffer is killed.
The command M-x kill-some-buffers asks about each buffer, one
by one. An answer of yes means to kill the buffer, just like
kill-buffer. This command ignores buffers whose names begin
with a space, which are used internally by Emacs.
The command M-x kill-matching-buffers prompts for a regular
expression and kills all buffers whose names match that expression.
See Syntax of Regular Expressions. Like kill-some-buffers, it asks for confirmation
before each kill. This command normally ignores buffers whose names
begin with a space, which are used internally by Emacs. To kill
internal buffers as well, call kill-matching-buffers with a
prefix argument. The command M-x kill-matching-buffers-no-ask
works like kill-matching-buffers, but doesn’t ask for
confirmation before killing each matching buffer.
The Buffer Menu feature is also convenient for killing various buffers. See Operating on Several Buffers.
If you want to do something special every time a buffer is killed, you
can add hook functions to the hook kill-buffer-hook (see Hooks).
If you run one Emacs session for a period of days, as many people do,
it can fill up with buffers that you used several days ago. The command
M-x clean-buffer-list is a convenient way to purge them; it kills
all the unmodified buffers that you have not used for a long time. An
ordinary buffer is killed if it has not been displayed for three days;
however, you can specify certain buffers that should never be killed
automatically, and others that should be killed if they have been unused
for a mere hour. These defaults, and other aspects of this command’s
behavior, can be controlled by customizing several options described
in the doc string of clean-buffer-list.
You can also have this buffer purging done for you, once a day,
by enabling Midnight mode. Midnight mode operates each day
at midnight; at that time, it runs clean-buffer-list, or
whichever functions you have placed in the normal hook
midnight-hook (see Hooks). To enable Midnight mode, use
the Customization buffer to set the variable midnight-mode to
t. See Easy Customization Interface.
Begin editing a buffer listing all Emacs buffers.
Similar, but do it in another window.
The Buffer Menu opened by C-x C-b (see Listing Existing Buffers) does not merely list buffers. It also allows you to perform various operations on buffers, through an interface similar to Dired (see Dired, the Directory Editor). You can save buffers, kill them (here called deleting them, for consistency with Dired), or display them.
To use the Buffer Menu, type C-x C-b and switch to the window displaying the *Buffer List* buffer. You can also type M-x buffer-menu to open the Buffer Menu in the selected window. Alternatively, the command M-x buffer-menu-other-window opens the Buffer Menu in another window, and selects that window.
The Buffer Menu is a read-only buffer, and can be changed only through the special commands described in this section. The usual cursor motion commands can be used in this buffer. The following commands apply to the buffer described on the current line:
Flag the buffer for deletion (killing), then move point to the next
line (Buffer-menu-delete). The deletion flag is indicated by
the character ‘D’ on the line, before the buffer name. The
deletion occurs only when you type the x command (see below).
Like d, but move point up instead of down
(Buffer-menu-delete-backwards).
Flag the buffer for saving (Buffer-menu-save). The save flag
is indicated by the character ‘S’ on the line, before the buffer
name. The saving occurs only when you type x. You may request
both saving and deletion for the same buffer.
Perform all flagged deletions and saves (Buffer-menu-execute).
Remove all flags from the current line, and move down
(Buffer-menu-unmark). With a prefix argument, moves up after
removing the flags.
Move to the previous line and remove all flags on that line
(Buffer-menu-backup-unmark).
Remove a particular flag from all lines
(Buffer-menu-unmark-all-buffers). This asks for a single
character, and unmarks buffers marked with that character; typing
RET removes all marks.
Remove all flags from all the lines
(Buffer-menu-unmark-all).
The commands for removing flags, d and C-d, accept a numeric argument as a repeat count.
The following commands operate immediately on the buffer listed on the current line. They also accept a numeric argument as a repeat count.
Mark the buffer as unmodified (Buffer-menu-not-modified).
See Commands for Saving Files.
Toggle the buffer’s read-only status
(Buffer-menu-toggle-read-only). See Miscellaneous Buffer Operations.
Visit the buffer as a tags table
(Buffer-menu-visit-tags-table). See Selecting a Tags Table.
The following commands are used to select another buffer or buffers:
Quit the Buffer Menu (quit-window). The most recent formerly
visible buffer is displayed in its place.
Select this line’s buffer, replacing the *Buffer List* buffer
in its window (Buffer-menu-this-window).
Select this line’s buffer in another window, as if by C-x 4 b,
leaving *Buffer List* visible
(Buffer-menu-other-window).
Display this line’s buffer in another window, without selecting it
(Buffer-menu-switch-other-window).
Select this line’s buffer in a full-frame window
(Buffer-menu-1-window).
Set up two windows on the current frame, with this line’s buffer
selected in one, and a previously current buffer (aside from
*Buffer List*) in the other (Buffer-menu-2-window).
Bury this line’s buffer (Buffer-menu-bury) (i.e., move it to
the end of the buffer list).
Mark this line’s buffer to be displayed in another window if you exit
with the v command (Buffer-menu-mark). The display flag
is indicated by the character ‘>’ at the beginning of the line.
(A single buffer may not have both deletion and display flags.)
Select this line’s buffer, and also display in other windows any
buffers flagged with the m command (Buffer-menu-select).
If you have not flagged any buffers, this command is equivalent to
1.
The following commands affect the entire buffer list:
Sort the Buffer Menu entries according to their values in the column
at point. With a numeric prefix argument n, sort according to
the n-th column (tabulated-list-sort).
Widen the current column width by n (the prefix numeric argument) characters.
Narrow the current column width by n (the prefix numeric argument) characters.
Delete, or reinsert, lines for non-file buffers
(Buffer-menu-toggle-files-only). This command toggles the
inclusion of such buffers in the buffer list.
Toggle display of internal buffers, those whose names begin with a space.
Normally, the buffer *Buffer List* is not updated
automatically when buffers are created and killed; its contents are
just text. If you have created, deleted or renamed buffers, the way
to update *Buffer List* to show what you have done is to type
g (revert-buffer). You can make this happen regularly
every auto-revert-interval seconds if you enable Auto Revert
mode in this buffer, as long as it is not marked modified. Global
Auto Revert mode applies to the *Buffer List* buffer only if
global-auto-revert-non-file-buffers is non-nil.
See global-auto-revert-non-file-buffers, for details.
An indirect buffer shares the text of some other buffer, which is called the base buffer of the indirect buffer. In some ways it is a buffer analogue of a symbolic link between files.
Create an indirect buffer named indirect-name with base buffer base-buffer.
Create an indirect buffer that is a twin copy of the current buffer.
Create an indirect buffer that is a twin copy of the current buffer, and
select it in another window (clone-indirect-buffer-other-window).
The text of the indirect buffer is always identical to the text of its base buffer; changes made by editing either one are visible immediately in the other. “Text” here includes both the characters and their text properties. But in all other respects, the indirect buffer and its base buffer are completely separate. They can have different names, different values of point, different narrowing, different markers, different overlays, different major modes, and different local variables.
An indirect buffer cannot visit a file, but its base buffer can. If you try to save the indirect buffer, that actually works by saving the base buffer. Killing the base buffer effectively kills the indirect buffer, but killing an indirect buffer has no effect on its base buffer.
One way to use indirect buffers is to display multiple views of an outline. See Viewing One Outline in Multiple Views.
A quick and handy way to make an indirect buffer is with the command
C-x 4 c (clone-indirect-buffer-other-window). It creates
and selects an indirect buffer whose base buffer is the current
buffer. With a numeric argument, it prompts for the name of the
indirect buffer; otherwise it uses the name of the current buffer,
with a ‘<n>’ suffix added.
The more general way to make an indirect buffer is with the command M-x make-indirect-buffer. It creates an indirect buffer named indirect-name from a buffer base-buffer, prompting for both using the minibuffer.
The functions that create indirect buffers run the hook
clone-indirect-buffer-hook after creating the indirect buffer.
When this hook runs, the newly created indirect buffer is the current
buffer.
Note: When a modification is made to the text of a buffer, the modification hooks are run only in the base buffer, because most of the functions on those hooks are not prepared to work correctly in indirect buffers. So if you need a modification hook function in an indirect buffer, you need to manually add that function to the hook in the base buffer and then make the function operate in the desired indirect buffer.
This section describes several modes and features that make it more convenient to switch between buffers.
When several buffers visit identically-named files, Emacs must give the buffers distinct names. The default method adds a suffix based on the names of the directories that contain the files. For example, if you visit files /foo/bar/mumble/name and /baz/quux/mumble/name at the same time, their buffers will be named ‘name<bar/mumble>’ and ‘name<quux/mumble>’, respectively. Emacs adds as many directory parts as are needed to make a unique name.
You can choose from several different styles for constructing unique
buffer names, by customizing the option uniquify-buffer-name-style.
The forward naming method includes part of the file’s
directory name at the beginning of the buffer name; using this method,
buffers visiting the files /u/rms/tmp/Makefile and
/usr/projects/zaphod/Makefile would be named
‘tmp/Makefile’ and ‘zaphod/Makefile’.
In contrast, the post-forward naming method would call the
buffers ‘Makefile|tmp’ and ‘Makefile|zaphod’. The default
method post-forward-angle-brackets is like post-forward,
except that it encloses the unique path in angle brackets. The
reverse naming method would call them ‘Makefile\tmp’ and
‘Makefile\zaphod’. The nontrivial difference between
post-forward and reverse occurs when just one directory
name is not enough to distinguish two files; then reverse puts
the directory names in reverse order, so that /top/middle/file
becomes ‘file\middle\top’, while post-forward puts them in
forward order after the file name, as in ‘file|top/middle’. If
uniquify-buffer-name-style is set to nil, the buffer
names simply get ‘<2>’, ‘<3>’, etc. appended.
The value of uniquify-buffer-name-style can be set to a
customized function with two arguments base and
extra-strings where base is a string and
extra-strings is a list of strings. For example the current
implementation for post-forward-angle-brackets could be:
(defun my-post-forward-angle-brackets (base extra-string) (concat base \"<\" (mapconcat #'identity extra-string \"/\") \">\"))
Which rule to follow for putting the directory names in the buffer name is not very important if you are going to look at the buffer names before you type one. But as an experienced user, if you know the rule, you won’t have to look. And then you may find that one rule or another is easier for you to remember and apply quickly.
Icomplete mode provides a convenient way to quickly select an element among the possible completions in a minibuffer. When enabled, typing in the minibuffer continuously displays a list of possible completions that match the string you have typed.
At any time, you can type C-j to select the first completion in the list. So the way to select a particular completion is to make it the first in the list. There are two ways to do this. You can type more of the completion name and thus narrow down the list, excluding unwanted completions above the desired one. Alternatively, you can use C-. and C-, to rotate the list until the desired buffer is first.
M-TAB will select the first completion in the list, like C-j but without exiting the minibuffer, so you can edit it further. This is typically used when entering a file name, where M-TAB can be used a few times to descend in the hierarchy of directories.
To enable Icomplete mode for the minibuffer, type M-x
icomplete-mode, or customize the variable icomplete-mode to
t (see Easy Customization Interface).
You can also additionally enable Icomplete mode for C-M-i (the
command completion-at-point) by customizing the variable
icomplete-in-buffer to t. For in-buffer completion, the
completion-auto-help variable controls when Icomplete mode’s
display of possible completions appears. The default value of
t means that the display of possible completions appears when
you first type C-M-i.
By default, when you press C-M-i, both Icomplete mode’s
in-buffer display of possible completions and the *Completions*
buffer appear. If you are using icomplete-in-buffer, then you
may wish to suppress this appearance of the *Completions*
buffer. To do that, add the following to your initialization file
(see The Emacs Initialization File):
(advice-add 'completion-at-point :after #'minibuffer-hide-completions)
An alternative to Icomplete mode is Fido mode. This is very similar
to Icomplete mode, but retains some functionality from a popular
extension called Ido mode (in fact the name is derived from “Fake
Ido”). Among other things, in Fido mode, C-s and C-r are
also used to rotate the completions list, C-k can be used to
delete files and kill buffers in-list. Another noteworthy aspect is
that flex is used as the default completion style
(see How Completion Alternatives Are Chosen). To change this, add the following to
your initialization file (see The Emacs Initialization File):
(defun my-icomplete-styles () (setq-local completion-styles '(initials flex))) (add-hook 'icomplete-minibuffer-setup-hook 'my-icomplete-styles)
To enable Fido mode, type M-x fido-mode, or customize
the variable fido-mode to t (see Easy Customization Interface).
Icomplete mode and Fido mode display the possible completions on the
same line as the prompt by default. To display the completion candidates
vertically under the prompt, type M-x icomplete-vertical-mode, or
customize the variable icomplete-vertical-mode to t
(see Easy Customization Interface).
Make a list of buffers similarly to M-x list-buffers but customizable.
Make a list of buffers and operate on them in Dired-like fashion.
M-x bs-show pops up a buffer list similar to the one normally
displayed by C-x C-b, but whose display you can customize in a
more flexible fashion. For example, you can specify the list of
buffer attributes to show, the minimum and maximum width of buffer
name column, a regexp for names of buffers that will never be shown
and those which will always be shown, etc. If you prefer
this to the usual buffer list, you can bind this command to C-x
C-b. To customize this buffer list, use the bs Custom group
(see Easy Customization Interface), or invoke bs-customize.
MSB global minor mode (“MSB” stands for “mouse select buffer”)
provides a different and customizable mouse buffer menu which you may
prefer. It replaces the mouse-buffer-menu commands, normally
bound to C-Down-mouse-1 and C-F10, with its own
commands, and also modifies the menu-bar buffer menu. You can
customize the menu in the msb Custom group.
IBuffer is a major mode for viewing a list of buffers and operating on them in a way analogous to that of Dired (see Dired, the Directory Editor), including filtering, marking, sorting in various ways, and acting on buffers.