Linux emacs command

On Unix-like operating systems, the emacs command launches Emacs, a comprehensive, customizable text editor, optimized for software development.

Syntax

emacs [ command-line switches ] [ files ... ]

Description

Emacs is a highly extensible, extremely customizable text editor originally written for the PDP-10 by Richard Stallman in the 1970s. It incorporates over 2000 commands, which can be extended using Emacs Lisp, and automated by user-created macros.

Emacs can do much more than simple insertion and deletion of text. It can control subprocesses, indent programs automatically, show multiple files at once, and more. Emacs editing commands operate in terms of characters, words, lines, sentences, paragraphs, and pages, and expressions and comments in various programming languages.

Emacs is called "self-documenting" because at any time you can use special commands, known as help commands, to find out what your options are, or to find out what any command does, or to find all the commands that pertain to a given topic.

Emacs is customizable in that you can easily alter its behavior in simple ways. For instance, if you use a programming language where comments start with '<**' and end with '**>', you can tell the Emacs comment manipulation commands to use those strings (see Comments). To use another example, you can rebind the basic cursor motion commands (up, down, left and right) to any keys on the keyboard that you find comfortable. See Customization.

Emacs is extensible in that you can go beyond simple customization and create entirely new commands. New commands are programs written in the Lisp language, which are run by Emacs's own Lisp interpreter. Existing commands can even be redefined in the middle of an editing session, without having to restart Emacs. Most of the editing commands in Emacs are written in Lisp; the few exceptions could be written in Lisp but use C instead for efficiency. Writing an extension is programming, but non-programmers can use it afterwards.

Getting help from within Emacs

Emacs has an extensive built-in help facility, but it assumes that you know how to manipulate Emacs windows and buffers. Ctrl-h or F1 enters the Help facility. "Help Tutorial" (Ctrl-h t) starts an interactive tutorial which can teach beginners the fundamentals of Emacs in a few minutes. "Help Apropos" (Ctrl-h a) helps you find a command given its functionality, "Help Character" (Ctrl-h c) describes a given character's effect, and "Help Function" (Ctrl-h f) describes a given Lisp function specified by name.

This documentation covers GNU Emacs, which includes special packages to handle mail reading (RMail) and mail sending (Mail); outline editing (Outline); compiling (Compile); running sub-shells within Emacs windows (Shell); running a Lisp read-eval-print loop (Lisp-Interaction-Mode), automated psychotherapy (seriously. It's called Doctor), and much more.

Options

General-interest Emacs options:

The following options are lisp-oriented (these options are processed in the order encountered):

The following options are useful when running Emacs as a batch editor:

Using Emacs with X

Emacs is designed to work well with the X Window System. If you run Emacs under X Windows, it creates its own window to be displayed. You probably want to run it as a background process so that you can still use your original window. Emacs can be started with the following X-related switches:

Configuring Emacs defaults for X

You can set X default values for your Emacs windows in your .Xresources file. Use the following format:

emacs.keyword:value

Where value specifies the default value of keyword. Emacs lets you set default values for the following keywords:

The organization of the Emacs screen

On a graphical display, such as on GNU/Linux using the X Window System, Emacs occupies a "graphical window". On a text terminal, Emacs occupies the entire terminal screen. We use the term frame to mean a graphical window or terminal screen occupied by Emacs. Emacs behaves very similarly on both kinds of frames. It normally starts out with only one frame, but you can create additional frames if you want (see Frames).

Each frame consists of several distinct regions. At the top of the frame is a menu bar, which allows you to access commands via a series of menus. On a graphical display, directly below the menu bar is a tool bar, a row of icons that perform editing commands if you click them. At the very bottom of the frame is an echo area, where informative messages are displayed and where you enter information when Emacs asks for it.

The main area of the frame, below the tool bar (if one exists) and above the echo area, is called the window. Graphical display systems commonly use the word "window" with a different meaning; but here we refer to those "graphical windows" as "frames".

An Emacs window is where the buffer (the text you are editing) is displayed. On a graphical display, the window possesses a scroll bar on one side, which can scroll through the buffer. The last line of the window is a mode line. This displays various information about what is going on in the buffer, such as whether there are unsaved changes, the editing modes that are in use, the current line number, and so forth.

When you start Emacs, there is normally only one window in the frame. However, you can subdivide this window horizontally or vertically to create multiple windows, each of which can independently display a buffer (see Windows).

At any time, one window is the selected window. On a graphical display, the selected window shows a more prominent cursor (usually solid and blinking); other windows show a less prominent cursor (usually a hollow box). On a text terminal, there is only one cursor, which is shown in the selected window. The buffer displayed in the selected window is called the current buffer, and it is where editing happens. Most Emacs commands implicitly apply to the current buffer; the text displayed in unselected windows is mostly visible for reference. If you use multiple frames on a graphical display, selecting a particular frame selects a window in that frame.

• Point: The place in the text where editing commands operate.
• Echo Area: Short messages appear at the bottom of the screen.
• Mode Line: Interpreting the mode line.
• Menu Bar: How to use the menu bar.

Point

The cursor in the selected window shows the location where most editing commands take effect, which is called point. Many Emacs commands move point to different places in the buffer; for example, you can place point by clicking mouse button 1 (normally the left button) at the desired location.

By default, the cursor in the selected window is drawn as a solid block and appears to be on a character, but think of point as between two characters; it is situated before the character under the cursor. For example, if your text looks like 'frob' with the cursor over the 'b', then point is between the 'o' and the 'b'. If you insert the character '!' at that position, the result is 'fro!b', with point between the '!' and the 'b'. Thus, the cursor remains over the 'b', as before.

If you are editing several files in Emacs, each in its own buffer, each buffer has its own value of point. A buffer that is not currently displayed remembers its value of point if you later display it again. Furthermore, if a buffer is displayed in multiple windows, each of those windows has its own value of point.

See Cursor Display, for options that control how Emacs displays the cursor.

The echo area

The line at the very bottom of the frame is the echo area. It is used to display small amounts of text for various purposes.

The echo area is so-named because one of the things it is used for is echoing, which means displaying the characters of a multi-character command as you type. Single-character commands are not echoed. Multi-character commands (see Keys) are echoed if you pause for more than a second in the middle of a command. Emacs then echoes all the characters of the command so far, to prompt you for the rest. Once echoing has started, the rest of the command echoes immediately as you type it. This behavior is designed to give confident users fast response, while giving hesitant users maximum feedback.

The echo area is also used to display an error message when a command cannot do its job. Error messages may be accompanied by beeping or by flashing the screen.

Some commands display informative messages in the echo area to tell you what the command has done, or to provide you with some specific information. These informative messages, unlike error messages, are not accompanied with a beep or flash. For example, C-x means the following: hold down Ctrl and type x, then let go of Ctrl and type =. This displays a message describing the character at point, its position in the buffer, and its current column in the window. Commands that take a long time often display messages ending in '...' while they are working (sometimes also indicating how much progress is made, as a percentage), and add 'done' when they are finished.

Informative echo area messages are saved in a special buffer named *Messages*. Buffers have not been explained yet; see Buffers for more information about them. If you miss a message that appeared briefly on the screen, you can switch to the *Messages* buffer to see it again. The *Messages* buffer is limited to a certain number of lines, specified by the variable message-log-max. See Variables for more information about them. Beyond this limit, one line is deleted from the beginning whenever a new message line is added at the end.

See Display Custom for options that control how Emacs uses the echo area.

The echo area is also used to display the minibuffer, a special window where you can input arguments to commands, such as the name of a file to be edited. When the minibuffer is in use, the text displayed in the echo area begins with a prompt string, and the active cursor appears in the minibuffer, which is temporarily considered the selected window. You can always get out of the minibuffer by typing C-g. See Minibuffer for more info.

The mode line

At the bottom of each window is a mode line, which describes what is going on in the current buffer. When there is only one window, the mode line appears right above the echo area; it is the next-to-last line in the frame. On a graphical display, the mode line is drawn with a 3D box appearance. Emacs also usually draws the mode line of the selected window with a different color than that of unselected windows, to make it stand out.

The text displayed in the mode line has the following format:

cs:ch-fr buf pos line (major minor)

On a text terminal, this text is followed by a series of dashes extending to the right edge of the window. These dashes are omitted on a graphical display.

The cs string and the colon character after it describe the character set and newline convention used for the current buffer. Normally, Emacs automatically handles these settings for you, but it is sometimes useful to have this information.

cs describes the character set of the text in the buffer (see Coding Systems). If it's a dash ('-'), that indicates no special character set handling (with the possible exception of end-of-line conventions, described in the next paragraph). '=' means no conversion whatsoever, and is usually used for files containing non-textual data. Other characters represent various coding systems—for example, '1' represents ISO Latin-1.

On a text terminal, cs is preceded by two additional characters that describe the coding systems for keyboard input and terminal output. If you are using an input method, cs is preceded by a string that identifies the input method (see Input Methods).

The character after cs is usually a colon. If a different string is displayed, that indicates a nontrivial end-of-line convention for encoding a file. Usually, lines of text are separated by newline characters in a file, but two other conventions are sometimes used. The MS-DOS convention uses a "carriage-return" character followed by a "linefeed" character; when editing such files, the colon changes to either a backslash ('\') or '(DOS)', depending on the operating system. Another convention, employed by older Macintosh systems, uses a "carriage-return" character instead of a newline; when editing such files, the colon changes to either a forward slash ('/') or '(Mac)'. On some systems, Emacs displays '(Unix)' instead of the colon for files that use newline as the line separator.

The next element on the mode line is the string indicated by ch. This shows two dashes ('--') if the buffer displayed in the window has the same contents as the corresponding file on the disk; i.e., if the buffer is "unmodified". If the buffer is modified, it shows two stars ('**'). For a read-only buffer, it shows '%*' if the buffer is modified, and '%%' otherwise.

The character after ch is normally a dash ('-'). However, if the default-directory for the current buffer is on a remote machine, '@' is displayed instead (see File Names).

fr gives the selected frame name (see Frames). It appears only on text terminals. The initial frame's name is 'F1'.

buf is the name of the buffer displayed in the window. Usually, this is the same as the name of a file you are editing. See Buffers.

pos tells you whether there is additional text above the top of the window, or below the bottom. If your buffer is small and it is visible in the window, pos is 'All'. Otherwise, it is 'Top' if you are looking at the beginning of the buffer, 'Bot' if you are looking at the end of the buffer, or 'nn%', where nn is the percentage of the buffer above the top of the window. With Size Indication mode, you can display the size of the buffer as well. See Optional Mode Line.

line is the character 'L' followed by the line number at point. You can display the current column number too, by turning on Column Number mode. See Optional Mode Line.

major is the name of the major mode used in the buffer. A major mode is a principal editing mode for the buffer, such as Text mode, Lisp mode, C mode, and so forth. See Major Modes. Some major modes display additional information after the major mode name. For example, Compilation buffers and Shell buffers display the status of the subprocess.

minor is a list of some of the enabled minor modes, which are optional editing modes that provide additional features on top of the major mode. See Minor Modes.

Some features are listed together with the minor modes whenever they are turned on, even though they are not really minor modes. 'Narrow' indicates the buffer being displayed has editing restricted to only a portion of its text (see Narrowing). 'Def' indicatest a keyboard macro is currently being defined (see Keyboard Macros).

Also, if Emacs is inside a recursive editing level, square brackets ('[...]') appear around the parentheses that surround the modes. If Emacs is in one recursive editing level within another, double square brackets appear. Since recursive editing levels affect Emacs globally, such square brackets appear in the mode line of every window. See Recursive Edit.

You can change the appearance of the mode line and the format of its contents; see Optional Mode Line. Also, the mode line is mouse-sensitive; clicking different parts of the mode line performs various commands. See Mode Line Mouse.

The menu bar

Each Emacs normally has a menu bar at the top which you can use to perform common operations. There's no need to list them here, as you can more easily see them yourself.

On a graphical display, you can use the mouse to choose a command from the menu bar. An arrow on the right edge of a menu item means it leads to a subsidiary menu, or "submenu". A '...' at the end of a menu item indicates the command prompts you for further input before it actually does anything.

Some of the commands in the menu bar have ordinary key bindings as well; if so, a key binding is shown in parentheses after the item itself. To view the full command name and documentation for a menu item, type C-h k, and then select the menu bar with the mouse in the usual way (see Key Help).

Instead of using the mouse, you can also invoke the first menu bar item by pressing F10 (to run the command menu-bar-open). You can then navigate the menus with the arrow keys. To activate a selected menu item, press enter; to cancel menu navigation, press escape. On a text terminal, you can use the menu bar by typing M-` or F10 (these run the command tmm-menubar). This lets you select a menu item with the keyboard. A provisional choice appears in the echo area. You can use the up and down arrow keys to move through the menu to different items, and then you can press enter to select the item. Each menu item is also designated by a letter or digit (usually the initial of some word in the item's name). This letter or digit is separated from the item name by '==>'. You can type the item's letter or digit to select the item.

Kinds of user input

Keyboard input in Emacs is based on a heavily-extended version of ASCII. Simple characters, like 'a', 'B', '3', '=', and the space character (denoted as <SPC>), are entered by typing the corresponding key. Control characters, such as <RET>, <TAB>, <DEL>, <ESC>, <F1>, <Home>, and <left>, are also entered this way, as are certain characters found on non-English keyboards (see International).

Emacs also recognizes control characters that are entered using modifier keys. Two commonly-used modifier keys are Control (usually labeled Ctrl), and Meta (usually labeled Alt). For example, Control-a is entered by holding down the Ctrl key while pressing a; this is referred to this as C-a for short. Similarly Meta-a, or M-a for short, is entered by holding down the Alt key and pressing a. Modifier keys can also be applied to non-alphanumerical characters, e.g., C-F1 or M-left.

You can also type Meta characters using two-character sequences starting with <ESC>. Thus, you can enter M-a by typing <ESC> a. You can enter C-M-a by typing <ESC> C-a. Unlike Meta, <ESC> is entered as a separate character. You don't hold down <ESC> while typing the next character; instead, press <ESC> and release it, then enter the next character. This feature is useful on certain text terminals where the Meta key does not function reliably.

On graphical displays, the window manager might block some keyboard inputs, including M-<TAB>, M-<SPC>, C-M-d and C-M-l. If you have this problem, you can either customize your window manager to not block those keys, or "rebind" the affected Emacs commands (see Customization).

Simple characters and control characters, and certain non-keyboard inputs such as mouse clicks, are collectively referred to as input events.

Keys

Some Emacs commands are invoked by only one input event; for example, C-f moves forward one character in the buffer. Other commands take two or more input events to invoke, such as C-x C-f and C-x 4 C-f.

A key sequence, or key for short, is a sequence of one or more input events that is meaningful as a unit. If a key sequence invokes a command, we call it a complete key; for example, C-f, C-x C-f and C-x 4 C-f are all complete keys. If a key sequence isn't long enough to invoke a command, we call it a prefix key; from the preceding example, we see that C-x and C-x 4 are prefix keys. Every key sequence is either a complete key or a prefix key.

A prefix key combines with the following input event to make a longer key sequence. For example, C-x is a prefix key, so typing C-x alone does not invoke a command; instead, Emacs waits for further input (if you pause for longer than a second, it echoes the C-x key to prompt for that input; see Echo Area). C-x combines with the next input event to make a two-event key sequence, which could itself be a prefix key (such as C-x 4), or a complete key (such as C-x C-f). There is no limit to the length of key sequences, but in practice they are seldom longer than three or four input events.

You can't add input events onto a complete key. For example, because C-f is a complete key, the two-event sequence C-f C-k is two key sequences, not one.

By default, the prefix keys in Emacs are C-c, C-h, C-x, C-x <RET>, C-x @, C-x a, C-x n, C-x r, C-x v, C-x 4, C-x 5, C-x 6, <ESC>, M-g, and M-o. (F1 and F2 are aliases for C-h and C-x 6.) This list is not cast in stone; if you customize Emacs, you can make new prefix keys. You could even eliminate some of the standard ones, though this is not recommended for most users; for example, if you remove the prefix definition of C-x 4, then C-x 4 C-f becomes an invalid key sequence. See Key Bindings. Typing the help character (C-h or F1) after a prefix key displays a list of the commands starting with that prefix. The sole exception to this rule is <ESC>: <ESC> C-h is equivalent to C-M-h, which does something else entirely. You can, however, use F1 to display a list of commands starting with <ESC>.

Keys and commands

Emacs does not assign meanings to keys directly. Instead, Emacs assigns meanings to named commands, and then gives keys their meanings by binding them to commands.

Every command has a name chosen by a programmer. The name is usually made of a few English words separated by dashes; for example, next-line or forward-word. Internally, each command is a special type of Lisp function, and the actions associated with the command are performed by running the function.

The bindings between keys and commands are recorded in tables called keymaps. See Keymaps. The description "C-n moves down vertically one line" glosses over a subtle distinction that is irrelevant in ordinary use, but vital for Emacs customization. The command next-line does a vertical move downward. C-n has this effect because it is bound to next-line. If you rebind C-n to the command forward-word, C-n moves forward one word instead.

In this manual, we often speak of keys like C-n as commands, even though strictly speaking the key is bound to a command. Usually we state the name of the command which really does the work in parentheses after mentioning the key that runs it. For example, we say that "The command C-n (next-line) moves point vertically down", meaning that the command next-line moves vertically down, and the key C-n is normally bound to it. Since we are discussing customization, we should tell you about variables. Often the description of a command says, "To change this, set the variable this-that." A variable is a name used to store a value. Most of the variables documented in this manual are meant for customization: some command or other part of Emacs examines the variable and behaves differently according to the value that you set. You can ignore the information about variables until you are interested in customizing them. Then read the basic information on variables (see Variables) and the information about specific variables makes sense.

Entering Emacs

The usual way to invoke Emacs is with the shell command emacs. From a terminal window running in the X Window System, you can run Emacs in the background with emacs &; this way, Emacs won't tie up the terminal window, so you can use it to run other shell commands.

When Emacs starts up, the initial frame displays a special buffer named '*GNU Emacs*'. This startup screen contains information about Emacs and links to common tasks that are useful for beginning users. For instance, activating the 'Emacs Tutorial' link opens the Emacs tutorial; this does the same thing as the command C-h t (help-with-tutorial). To activate a link, either move point onto it and type <RET>, or click it with mouse-1 (the left mouse button).

Using a command line argument, you can tell Emacs to visit one or more files as soon as it starts up. For example, emacs foo.txt starts Emacs with a buffer displaying the contents of the file 'foo.txt'. This feature exists mainly for compatibility with other editors, which are designed to be launched from the shell for short editing sessions. If you call Emacs this way, the initial frame is split into two windows: one showing the specified file, and the other showing the startup screen. See Windows.

Generally, it is unnecessary and wasteful to start Emacs afresh each time you want to edit a file. The recommended way to use Emacs is to start it only once, only after you log in, and do all your editing in the same Emacs session. See Files, for information on visiting more than one file. If you use Emacs this way, the Emacs session accumulates valuable context, such as the kill ring, registers, undo history, and mark ring data, which together make editing more convenient. These features are described later in the manual.

To edit a file from another program while Emacs is running, you can use the emacsclient helper program to open a file in the existing Emacs session. See Emacs Server.

Emacs accepts other command line arguments that tell it to load certain Lisp files, where to put the initial frame, and so forth. See Emacs Invocation.

If the variable inhibit-startup-screen is set to anything, Emacs does not display the startup screen. In that case, if one or more files were specified on the command line, Emacs displays those files; otherwise, it displays a buffer named *scratch*, which can evaluate Emacs Lisp expressions interactively. See Lisp Interaction. You can set the variable inhibit-startup-screen using the Customize facility (see Easy Customization), or by editing your initialization file (see Init File).

You can also force Emacs to display a file or directory at startup by setting the variable initial-buffer-choice to a value. In that case, even if you specify one or more files on the command line, Emacs opens but does not display them. The value of initial-buffer-choice should be the name of the desired file or directory.

Exiting Emacs

C-x C-c

Kills Emacs (save-buffers-kill-terminal).

C-z

On a text terminal, this suspends Emacs; on a graphical display, "minimize" the selected frame (suspend-emacs).

Killing Emacs means terminating the Emacs program. To do this, type C-x C-c (save-buffers-kill-terminal). A two-character key sequence is used to make it harder to type by accident. If there are any modified file-visiting buffers when you type C-x C-c, Emacs first offers to save these buffers. If you do not save them all, it asks for confirmation again, as the unsaved changes are lost. Emacs also asks for confirmation if any subprocesses are still running, as killing Emacs also kills the subprocesses (see Shell).

C-x C-c behaves specially if you are using Emacs as a server. If you type it from a "client frame", it closes the client connection. See Emacs Server.

Emacs can, optionally, record certain session information when you kill it, such as the files you were visiting at the time. This information is then available the next time you start Emacs. See Saving Emacs Sessions. If the value of the variable confirm-kill-emacs is non-nil, C-x C-c assumes that its value is a predicate function, and calls that function. If the result of the function call is not nil, the session is killed, otherwise Emacs continues to run. One convenient function to use as the value of confirm-kill-emacs is the function yes-or-no-p. The default value of confirm-kill-emacs is nil.

To kill Emacs without being prompted about saving, type M-x kill-emacs.

C-z runs the command suspend-frame. On a graphical display, this command minimizes (or iconifies) the selected Emacs frame, hiding it in a way that lets you bring it back later (exactly how this hiding occurs depends on the window system). On a text terminal, the C-z command suspends Emacs, stopping the program temporarily and returning control to the parent process (usually a shell); in most shells, you can resume Emacs after suspending it with the shell command %emacs.

Text terminals usually listen for certain special characters whose meaning is to kill or suspend the program you are running. This terminal feature is turned off while you are in Emacs. The meanings of C-z and C-x C-c as keys in Emacs were inspired by the use of C-z and C-c on several operating systems as the characters for stopping or killing a program, but that is their only relationship with the operating system. You can customize these keys to run any commands of your choice (see Keymaps).

Basic editing commands

The following sections explain the basics of how to enter text, make corrections, and save the text in a file. If this material is new to you, it's suggested you first run the Emacs learn-by-doing tutorial, by typing C-h t (help-with-tutorial).

• Inserting Text: Inserting text by typing it.
• Moving Point: Moving the cursor to the place where you want to change something.
• Erasing: Deleting and killing text.
• Basic Undo: Undoing recent changes in the text.
• Files: Visiting, creating, and saving files.
• Help: Asking what a character does.
• Blank Lines: Making and deleting blank lines.
• Continuation Lines: How Emacs displays lines too wide for the screen.
• Position Info: What line, row, or column is point on?
• Arguments: Numeric arguments for repeating a command N times.
• Repeating: Repeating the previous command quickly.

Inserting text

You can insert an ordinary graphic character (e.g., 'a', 'B', '3', and '=') by typing the associated key. This adds the character to the buffer at point. Insertion moves point forward, so that point remains only after the inserted text. See Point.

To end a line and start a new one, type <RET> (newline). The <RET> key may be labeled Return or Enter on your keyboard, but Emacs refers to it as <RET>. This command inserts a newline character into the buffer. If point is at the end of the line, the effect is to create a new blank line after it; if point is in the middle of a line, the line is split at that position.

You can change the way Emacs handles text insertion by turning on minor modes. For instance, the minor mode called Auto Fill mode splits lines automatically when they get too long (see Filling). The minor mode called Overwrite mode causes inserted characters to replace (overwrite) existing text, instead of shoving it to the right. See Minor Modes.

Only graphic characters can be inserted by typing the associated key; other keys act as editing commands and do not insert themselves. For instance, <DEL> runs the command delete-backward-char by default (some modes bind it to a different command); it does not insert a literal '<DEL>' character (ASCII character code 127).

To insert a non-graphic character, or a character your keyboard does not support, first quote it by typing C-q (quoted-insert). There are two ways to use C-q:

• C-q followed by any non-graphic character (even C-g) inserts that character. For instance, C-q <DEL> inserts a literal '<DEL>' character.
• C-q followed by a sequence of octal digits inserts the character with the specified octal character code. You can use any number of octal digits; any non-digit terminates the sequence. If the terminating character is <RET>, that <RET> serves only to terminate the sequence. Any other non-digit terminates the sequence and then acts as normal input; thus, C-q 1 0 1 B inserts 'AB'. The use of octal sequences is disabled in ordinary non-binary Overwrite mode, to give you a convenient way to insert a digit instead of overwriting with it.

To use decimal or hexadecimal instead of octal, set the variable read-quoted-char-radix to 10 or 16. If the radix is 16, the letters a to f serve as part of a character code, only like digits. Case is ignored.

Alternatively, you can use the command C-x 8 <RET> (insert-char). This prompts for the Unicode name or code-point of a character, using the minibuffer. If you enter a name, the command provides completion (see Completion). If you enter a code-point, it should be as a hexadecimal number (the convention for Unicode), or a number with a specified radix, e.g., #o23072 (octal). The command then inserts the corresponding character into the buffer. For example, both of the following insert the infinity sign (Unicode code-point U+221E):

• C-x 8 <RET> infinity <RET>
• C-x 8 <RET> 221e <RET>

A numeric argument to C-q or C-x 8 <RET> specifies how many copies of the character to insert (see Arguments).

Changing the location of point

To do more than insert characters, you have to know how to move point (see Point). The keyboard commands C-f, C-b, C-n, and C-p move point to the right, left, down, and up, respectively. You can also move point using the arrow keys; however, many Emacs users find that it is slower to use the arrow keys than the control keys, because you need to move your hand to the area of the keyboard where those keys are located.

You can also click the left mouse button to move point to the position clicked. Emacs also provides a variety of additional keyboard commands that move point in more sophisticated ways.

When a line of text in the buffer is longer than the width of the window, Emacs usually displays it on two or more screen lines. For convenience, C-n and C-p move point by screen lines, as do the equivalent keys down and up. You can force these commands to move according to logical lines (i.e., according to the text lines in the buffer) by setting the variable line-move-visual to nil; if a logical line occupies multiple screen lines, the cursor then skips over the additional screen lines. For details, see Continuation Lines. See Variables, for how to set variables such as line-move-visual.

Unlike C-n and C-p, most of the Emacs commands that work on lines work on logical lines. For instance, C-a (move-beginning-of-line) and C-e (move-end-of-line) respectively move to the beginning and end of the logical line. Whenever we encounter commands that work on-screen lines, such as C-n and C-p, we point these out.

When line-move-visual is nil, you can also set the variable track-eol to a non-nil value. Then C-n and C-p, when starting at the end of the logical line, move to the end of the next logical line. Normally, track-eol is nil.

C-n normally stops at the end of the buffer when you use it on the last line in the buffer. However, if you set the variable next-line-add-newlines to a non-nil value, C-n on the last line of a buffer creates an additional line at the end and moves down into it.

Erasing text

The <DEL> (delete-backward-char) command removes the character before point, moving the cursor and the characters after it backwards. If point was at the beginning of a line, this deletes the preceding newline, joining this line to the previous one.

If, however, the region is active, <DEL> instead deletes the text in the region. See Mark, for a description of the region.

On most keyboards, <DEL> is labeled Backspace, but it is referred to as <DEL> in this manual. Do not confuse <DEL> with the Delete key; Delete is discussed momentarily. On some text terminals, Emacs may not recognize the <DEL> key properly.

The delete (delete-forward-char) command deletes in the "opposite direction": it deletes the character after point, i.e., the character under the cursor. If point was at the end of a line, this joins the following line onto this one. Like <DEL>, it deletes the text in the region if the region is active (see Mark).

C-d (delete-char) deletes the character after point, similar to delete, but regardless of whether the region is active.

See Deletion, for more detailed information about the above deletion commands.

C-k (kill-line) erases ("kills") a line at a time. If you type C-k at the beginning or middle of a line, it kills all the text up to the end of the line. If you type C-k at the end of a line, it joins that line with the following line. See Killing, for more information about C-k and related commands.

Undoing changes

Emacs records a list of changes made in the buffer text, so you can undo recent changes. This is done using the undo command, which is bound to C-/ (also C-x u and C-_). Normally, this command undoes the last change, moving point back to where it was before the change. The undo command applies only to changes in the buffer; you can't use it to undo cursor motion.

Although each editing command usually makes a separate entry in the undo records, very simple commands may be grouped together. Sometimes, an entry may cover only part of a complex command.

If you repeat C-/ (or its aliases), each repetition undoes another, earlier change, back to the limit of the undo information available. If all recorded changes have already been undone, the undo command displays an error message and does nothing.

To learn more about the undo command, see Undo.

Files

Text that you insert in an Emacs buffer lasts only as long as the Emacs session. To keep any text permanently, you must put it in a file.

Suppose there is a file named test.emacs in your home directory. To begin editing this file in Emacs, type

C-x C-f test.emacs <RET>

Here the file name is given as an argument to the command C-x C-f (find-file). That command uses the minibuffer to read the argument, and you type <RET> (press the Return/Enter key) to terminate the argument. See Minibuffer for more information on the minibuffer.

Emacs obeys this command by visiting the file: it creates a buffer, copies the contents of the file into the buffer, and then displays the buffer for editing. If you alter the text, you can save the new text in the file by typing C-x C-s (save-buffer). This copies the altered buffer contents back into the file test.emacs, making them permanent. Until you save, the changed text exists only inside Emacs, and the file test.emacs is unaltered.

To create a file, only visit it with C-x C-f as if it already existed. This creates an empty buffer, where you can insert the text you want to put in the file. Emacs actually creates the file the first time you save this buffer with C-x C-s.

To learn more in depth about how Emacs uses files, see Files.

Help

If you forget what a key does, you can find out by typing C-h k (describe-key), followed by the key of interest; for example, C-h k C-n tells you what C-n does.

The prefix key C-h stands for "help". The key F1 serves as an alias for C-h. Apart from C-h k, there are other help commands providing different kinds of help.

For more information in depth about Emacs Help, see Help.

Blank lines

Here are special commands and techniques for inserting and deleting blank lines:

We have seen how <RET> (newline) starts a new line of text. However, it may be easier to see what you are doing if you first make a blank line and then insert the desired text into it. This is easy to do using the key C-o (open-line), which inserts a newline after point but leaves point in front of the newline. After C-o, type the text for the new line.

You can make several blank lines by typing C-o several times, or by giving it a numeric argument specifying how many blank lines to make. See Arguments, for how. If you have a fill prefix, the C-o command inserts the fill prefix on the new line, if typed at the beginning of a line. See Fill Prefix.

The easy way to get rid of extra blank lines is with the command C-x C-o (delete-blank-lines). If point lies within a run of several blank lines, C-x C-o deletes all but one of them. If point is on a single blank line, C-x C-o deletes it. If point is on a nonblank line, C-x C-o deletes all following blank lines, if any exists.

Continuation lines

Sometimes, a line of text in the buffer (a logical line) is too long to fit in the window, and Emacs displays it as two or more screen lines. This is called line wrapping or continuation, and the long logical line is called a continued line. On a graphical display, Emacs indicates line wrapping with small bent arrows in the left and right window fringes. On a text terminal, Emacs indicates line wrapping by displaying a '\' character at the right margin.

Most commands that act on lines act on logical lines, not screen lines. For instance, C-k kills a logical line. As described earlier, C-n (next-line) and C-p (previous-line) are special exceptions: they move point down and up, respectively, by one screen line (see Moving Point).

Emacs can optionally truncate long logical lines instead of continuing them. This means that every logical line occupies a single screen line; if it's longer than the width of the window, the rest of the line is not displayed. On a graphical display, a truncated line is indicated by a small straight arrow in the right fringe; on a text terminal, it is indicated by a '$' character in the right margin. See Line Truncation.

By default, continued lines are wrapped at the right window edge. Since the wrapping may occur in the middle of a word, continued lines can be difficult to read. The usual solution is to break your lines before they get too long, by inserting newlines. If you prefer, you can make Emacs insert a newline automatically when a line gets too long, using Auto Fill mode. See Filling.

Sometimes, you may need to edit files containing many long logical lines, and it may not be practical to break them all up by adding newlines. In that case, you can use Visual Line mode, which enables word wrapping: instead of wrapping long lines exactly at the right window edge, Emacs wraps them at the word boundaries (i.e., space or tab characters) nearest to the right window edge. Visual Line mode also redefines editing commands such as C-a, C-n, and C-k to operate on-screen lines rather than logical lines. See Visual Line Mode.

Cursor position information

Here are commands to get information about the size and position of parts of the buffer, and to count words and lines.

M-x what-line displays the current line number in the echo area. This command is usually redundant, because the current line number is shown in the mode line (see Mode Line). However, if you narrow the buffer, the mode line shows the line number relative to the accessible portion (see Narrowing). By contrast, what-line displays both the line number relative to the narrowed region and the line number relative to the whole buffer.

M-= (count-words-region) displays a message reporting the number of lines, words, and characters in the region (see Mark, for an explanation of the region). With a prefix argument, C-u M-=, the command displays a count for the entire buffer.

The command M-x count-words does the same job, but with a different calling convention. It displays a count for the region if the region is active, and for the buffer otherwise.

The command C-x = (what-cursor-position) shows information about the current cursor position and the buffer contents at that position. It displays a line in the echo area that looks like this:

Char: c (99, #o143, #x63) point=28062 of 36168 (78%) column=53

After 'Char:', this shows the character in the buffer at point. The text inside the parenthesis shows the corresponding decimal, octal and hex character codes; for more information about how C-x = displays character information, see International Chars. After 'point=' is the position of point as a character count (the first character in the buffer is position 1, the second character is position 2, and so on). The number after that is the total number of characters in the buffer, and the number in parenthesis expresses the position as a percentage of the total. After 'column=' is the horizontal position of point, in columns counting from the left edge of the window.

If the buffer is narrowed, making some of the text at the beginning and the end temporarily inaccessible, C-x = displays additional text describing the currently accessible range. For example, it might display this:

Char: C (67, #o103, #x43) point=252 of 889 (28%) <231-599> column=0

Where the two extra numbers give the smallest and largest character position that point is allowed to assume. The characters between those two positions are the accessible ones. See Narrowing.

Numeric arguments

In the terminology of mathematics and computing, argument means "data provided to a function or operation". You can give any Emacs command a numeric argument (also called a prefix argument). Some commands interpret the argument as a repetition count. For example, giving C-f an argument of ten causes it to move point forward by ten characters instead of one. With these commands, no argument is equivalent to an argument of one, and negative arguments cause them to move or act in the opposite direction.

The easiest way to specify a numeric argument is to type a digit and/or a minus sign while holding down the META key. For example,

M-5 C-n

Moves down five lines. The keys M-1, M-2, and so on, and M--, are bound to commands (digit-argument and negative-argument) that set up an argument for the next command. Meta-- without digits normally means −1.

If you enter more than one digit, you need not hold down the META key for the second and subsequent digits. Thus, to move down fifty lines, type

M-5 0 C-n

Note that this does not insert five copies of '0' and move down one line, as you might expect. The '0' is treated as part of the prefix argument.

What if you do want to insert five copies of '0'? Type M-5 C-u 0. Here, C-u "terminates" the prefix argument, so that the next keystroke begins the command that you want to execute. Note that this meaning of C-u applies only to this case. For the usual role of C-u, see below. Instead of typing M-1, M-2, and so on, another way to specify a numeric argument is to type C-u (universal-argument) followed by some digits, or (for a negative argument) a minus sign followed by digits. A minus sign without digits normally means −1.

C-u alone has the special meaning of "four times": it multiplies the argument for the next command by four. C-u C-u multiplies it by sixteen. Thus, C-u C-u C-f moves forward sixteen characters. Other useful combinations are C-u C-n, C-u C-u C-n (move down a good fraction of a screen), C-u C-u C-o (make "a lot" of blank lines), and C-u C-k (kill four lines).

You can use a numeric argument before a self-inserting character to insert multiple copies of it. This is straightforward when the character is not a digit; for example, C-u 6 4 a inserts 64 copies of the character 'a'. But this does not work for inserting digits; C-u 6 4 1 specifies an argument of 641. You can separate the argument from the digit to insert with another C-u; for example, C-u 6 4 C-u 1 does insert 64 copies of the character '1'.

Some commands care whether there is an argument, but ignore its value. For example, the command M-q (fill-paragraph) fills text; with an argument, it onlyifies the text as well. (See Filling, for more information on M-q.) For these commands, it is enough to specify the argument with a single C-u.

Some commands use the value of the argument as a repeat count, but do something special when there is no argument. For example, the command C-k (kill-line) with argument n kills n lines, including their terminating newlines. But C-k with no argument is special: it kills the text up to the next newline, or, if point is right at the end of the line, it kills the newline itself. Thus, two C-k commands with no arguments can kill a nonblank line, only like C-k with an argument of one. See Killing, for more information on C-k.

A few commands treat a plain C-u differently from an ordinary argument. A few others may treat an argument of only a minus sign differently from an argument of −1. These unusual cases are described when they come up; they exist to make an individual command more convenient, and they are documented in that command's documentation string.

We use the term prefix argument to emphasize that you type such arguments before the command, and to distinguish them from minibuffer arguments (see Minibuffer), which are entered after invoking the command.

Repeating a command

Many simple commands, such as those invoked with a single key or with M-x command-name <RET>, can be repeated by invoking them with a numeric argument that serves as a repeat count (see Arguments). However, if the command you want to repeat prompts for input, or uses a numeric argument in another way, that method won't work.

The command C-x z (repeat) provides another way to repeat an Emacs command many times. This command repeats the previous Emacs command, whatever that was. Repeating a command uses the same arguments that were used before; it does not read new arguments each time.

To repeat the command more than once, type additional z's: each z repeats the command one more time. Repetition ends when you type a character other than z, or press a mouse button.

For example, suppose you type C-u 2 0 C-d to delete 20 characters. You can repeat that command (including its argument) three additional times, to delete a total of 80 characters, by typing C-x z z z. The first C-x z repeats the command once, and each subsequent z repeats it once again.

The minibuffer

The minibuffer is where Emacs commands read complicated arguments, such as file names, buffer names, Emacs command names, or Lisp expressions. We call it the "minibuffer" because it's a special-purpose buffer with a small amount of screen space. You can use the usual Emacs editing commands in the minibuffer to edit the argument text.

• Basic Minibuffer: Basic usage of the minibuffer.
• Minibuffer File: Entering file names with the minibuffer.
• Minibuffer Edit: How to edit in the minibuffer.
• Completion: An abbreviation facility for minibuffer input.
• Minibuffer History: Reusing recent minibuffer arguments.
• Repetition: Re-executing commands that used the minibuffer.
• Passwords: Entering passwords in the echo area.
• Yes or No Prompts: Replying yes or no in the echo area.

Using the minibuffer

When the minibuffer is in use, it appears in the echo area, with a cursor. The minibuffer starts with a prompt, usually ending with a colon. The prompt states what kind of input is expected, and how it is used. The prompt is highlighted using the minibuffer-prompt face (see Faces).

The simplest way to enter a minibuffer argument is to type the text, then <RET> to submit the argument and exit the minibuffer. Alternatively, you can type C-g to exit the minibuffer by canceling the command asking for the argument (see Quitting).

Sometimes, the prompt shows a default argument, inside parentheses before the colon. This default is used as the argument if you only type <RET>. For example, commands that read buffer names usually show a buffer name as the default; you can type <RET> to operate on that default buffer.

If you enable Minibuffer Electric Default mode, a global minor mode, Emacs hides the default argument as soon as you modify the contents of the minibuffer (since typing <RET> would no longer submit that default). If you ever bring back the original minibuffer text, the prompt again shows the default. Further, if you change the variable minibuffer-eldef-shorten-default to a non-nil value, the default argument is displayed as '[default]' instead of '(default default)', saving some screen space. To enable this minor mode, type M-x minibuffer-electric-default-mode.

Since the minibuffer appears in the echo area, it can conflict with other uses of the echo area. If an error message or an informative message is emitted while the minibuffer is active, the message hides the minibuffer for a few seconds, or until you type something; then the minibuffer comes back. While the minibuffer is in use, keystrokes do not echo.

Minibuffers for file names

Commands such as C-x C-f (find-file) use the minibuffer to read a file name argument (see Basic Files). When the minibuffer is used to read a file name, it often starts out with some initial text ending in a slash. This is the default directory. For example, it may start out like this:

Find file: /u2/emacs/src/

Here, 'Find file: ' is the prompt and '/u2/emacs/src/' is the default directory. If you now type buffer.c as input, that specifies the file /u2/emacs/src/buffer.c. See File Names, for information about the default directory.

You can specify the parent directory with ..: /a/b/../foo.el is equivalent to /a/foo.el. Alternatively, you can use M-<DEL> to kill directory names backwards (see Words).

To specify a file in a completely different directory, you can kill the entire default with C-a C-k (see Minibuffer Edit). Alternatively, you can ignore the default, and enter an absolute file name starting with a slash or a tilde after the default directory. For example, you can specify /etc/termcap as follows:

Find file: /u2/emacs/src//etc/termcap

Emacs interprets a double slash as "ignore everything before the second slash in the pair". In the example above, /u2/emacs/src/ is ignored, so the argument you supplied is /etc/termcap. The ignored part of the file name is dimmed if the terminal allows it. (To disable this dimming, turn off File Name Shadow mode with the command M-x file-name-shadow-mode.)

Emacs interprets ~/ as your home directory. Thus, ~/foo/bar.txt specifies a file named bar.txt, inside a directory named foo, which is in turn located in your home directory. Also, ~user-id/ means the home directory of a user whose login name is user-id. Any leading directory name in front of the ~ is ignored: thus, /u2/emacs/~/foo/bar.txt is equivalent to ~/foo/bar.txt.

On Microsoft Windows and MS-DOS systems, where a user doesn't always have a home directory, Emacs uses several alternatives. For Microsoft Windows, see Windows HOME. On these systems, the ~user-id/ construct is supported only for the current user, i.e., only if user-id is the current user's login name.

To prevent Emacs from inserting the default directory when reading file names, change the variable insert-default-directory to nil. In that case, the minibuffer starts out empty. Nonetheless, relative file name arguments are still interpreted based on the same default directory.

You can also enter remote file names in the minibuffer. See Remote Files.

Editing in the minibuffer

The minibuffer is an Emacs buffer, albeit a peculiar one, and the usual Emacs commands are available for editing the argument text. (The prompt, however, is read-only, and cannot be changed.)

Since <RET> in the minibuffer submits the argument, you can't use it to insert a newline. You can do that with C-q C-j, which inserts a C-j control character, which is formally equivalent to a newline character (see Inserting Text). Alternatively, you can use the C-o (open-line) command (see Blank Lines).

Inside a minibuffer, the keys <TAB>, <SPC>, and ? are often bound to completion commands, which allow you to easily fill in the desired text without typing all of it. See Completion. As with <RET>, you can use C-q to insert a <TAB>, <SPC>, or '?' character.

For convenience, C-a (move-beginning-of-line) in a minibuffer moves point to the beginning of the argument text, not the beginning of the prompt. For example, this allows you to erase the entire argument with C-a C-k.

When the minibuffer is active, the echo area is treated much like an ordinary Emacs window. For instance, you can switch to another window (with C-x o), edit text there, then return to the minibuffer window to finish the argument. You can even kill text in another window, return to the minibuffer window, and yank the text into the argument. There are some restrictions on the minibuffer window, however: for instance, you cannot split it. See Windows.

Normally, the minibuffer window occupies a single screen line. However, if you add two or more lines' worth of text into the minibuffer, it expands automatically to accommodate the text. The variable resize-mini-windows controls the resizing of the minibuffer. The default value is grow-only, which means the behavior we have only described. If the value is t, the minibuffer window also shrinks automatically if you remove some lines of text from the minibuffer, down to a minimum of one screen line. If the value is nil, the minibuffer window never changes size automatically, but you can use the usual window-resizing commands on it (see Windows).

The variable max-mini-window-height controls the maximum height for resizing the minibuffer window. A floating-point number specifies a fraction of the frame's height; an integer specifies the maximum number of lines; nil means do not resize the minibuffer window automatically. The default value is 0.25.

The C-M-v command in the minibuffer scrolls the help text from commands that display help text of any sort in another window. You can also scroll the help text with M-prior and M-next (or, equivalently, M-PageUp and M-PageDown). This is especially useful with long lists of possible completions. See Other Window.

Emacs normally disallows most commands that use the minibuffer while the minibuffer is active. To allow such commands in the minibuffer, set the variable enable-recursive-minibuffers to t.

When not active, the minibuffer is in minibuffer-inactive-mode, and clicking Mouse-1 there shows the *Messages* buffer. If you use a dedicated frame for minibuffers, Emacs also recognizes certain keys there, for example n to make a new frame.

Completion

You can often use a feature called completion to help enter arguments. This means that after you type part of the argument, Emacs can fill in the rest, or some of it, based on what was typed so far.

When completion is available, certain keys (usually <TAB>, <RET>, and <SPC>) are rebound in the minibuffer to special completion commands (see Completion Commands). These commands attempt to complete the text in the minibuffer, based on a set of completion alternatives provided by the command that requested the argument. You can usually type ? to see a list of completion alternatives.

Although completion is usually done in the minibuffer, the feature is sometimes available in ordinary buffers too. See Symbol Completion.

• Completion Example: Examples of using completion.
• Completion Commands: A list of completion commands.
• Completion Exit: Completion and minibuffer text submission.
• Completion Styles: How completion matches are chosen.
• Options: Options for completion.

Completion example

A simple example: M-x uses the minibuffer to read the name of a command, so completion works by matching the minibuffer text against the names of existing Emacs commands. Suppose you want to run the command auto-fill-mode. You can do that by typing M-x auto-fill-mode <RET>, but it is easier to use completion.

If you type M-x a u <TAB>, the <TAB> looks for completion alternatives (in this case, command names) that start with 'au'. There's several, including auto-fill-mode and autoconf-mode, but they all begin with auto, so the 'au' in the minibuffer completes to 'auto'. (More commands may be defined in your Emacs session. For example, if a command called authorize-me was defined, Emacs could only complete as far as 'aut'.)

If you type <TAB> again immediately, it cannot determine the next character; it could be '-', 'a', or 'c'. So it does not add any characters; instead, <TAB> displays a list of all possible completions in another window.

Next, type -f. The minibuffer now contains 'auto-f', and the only command name that starts with this is auto-fill-mode. If you now type <TAB>, completion fills in the rest of the argument 'auto-fill-mode' into the minibuffer.

Hence, typing only a u <TAB> - f <TAB> allows you to enter 'auto-fill-mode'.

Completion commands

Here is a list of the completion commands defined in the minibuffer when completion is allowed.

<TAB> (minibuffer-complete) is the most fundamental completion command. It searches for all possible completions that match the existing minibuffer text, and attempts to complete as much as it can. See Completion Styles, for how completion alternatives are chosen.

<SPC> (minibuffer-complete-word) completes like <TAB>, but only up to the next hyphen or space. If you have 'auto-f' in the minibuffer and type <SPC>, it finds that the completion is 'auto-fill-mode', but it only inserts 'ill-', giving 'auto-fill-'. Another <SPC> at this point completes to 'auto-fill-mode'.

If <TAB> or <SPC> cannot complete, it displays a list of matching completion alternatives (if there are any) in another window. You can display the same list with ? (minibuffer-completion-help). The following commands can be used with the completion list:

Completion exit

When a command reads an argument using the minibuffer with completion, it also controls what happens when you type <RET> (minibuffer-complete-and-exit) to submit the argument. There are four types of behavior:

• Strict completion accepts only exact completion matches. Typing <RET> exits the minibuffer only if the minibuffer text is an exact match, or completes to one. Otherwise, Emacs refuses to exit the minibuffer; instead it tries to complete, and if no completion can be done it momentarily displays '[No match]' after the minibuffer text. (You can still leave the minibuffer by typing C-g to cancel the command.)
• An example of a command that uses this behavior is M-x, as it is meaningless for it to accept a non-existent command name.
• Cautious completion is like strict completion, except <RET> exits only if the text is already an exact match. If the text completes to an exact match, <RET> performs that completion but does not exit yet; you must type a second <RET> to exit. Cautious completion is used for reading file names for files that must already exist, for example.
• Permissive completion allows any input; the completion candidates are only suggestions. Typing <RET> does not complete, it only submits the argument as you have entered it.
• Permissive completion with confirmation is like permissive completion, with an exception: if you typed <TAB> and this completed the text up to some intermediate state (i.e., one that is not yet an exact completion match), typing <RET> right afterward does not submit the argument. Instead, Emacs asks for confirmation by momentarily displaying '[Confirm]' after the text; type <RET> again to confirm and submit the text. This catches a common mistake, where one types <RET> before realizing that <TAB> did not complete as far as desired.
• You can tweak the confirmation behavior by customizing the variable confirm-nonexistent-file-or-buffer. The default value, after-completion, gives the behavior we have only described. If you change it to nil, Emacs does not ask for confirmation, falling back on permissive completion. If you change it to any other non-nil value, Emacs asks for confirmation whether or not the preceding command was <TAB>.

This behavior is used by most commands that read file names, like C-x C-f, and commands that read buffer names, like C-x b.

How completion alternatives are chosen

Completion commands work by narrowing a large list of possible completion alternatives to a smaller subset that "matches" what you have typed in the minibuffer. In Completion Example, we gave a simple example of such matching. The procedure of determining what constitutes a "match" is quite intricate. Emacs attempts to offer plausible completions under most circumstances.

Emacs performs completion using one or more completion styles—sets of criteria for matching minibuffer text to completion alternatives. During completion, Emacs tries each completion style in turn. If a style yields one or more matches, that is used as the list of completion alternatives. If a style produces no matches, Emacs falls back on the next style.

The list variable completion-styles specifies the completion styles to use. Each list element is the name of a completion style (a Lisp symbol). The default completion styles are (in order):

The following additional completion styles are also defined, and you can add them to completion-styles if you want (see Customization):

There is also a very simple completion style called emacs21. In this style, if the text in the minibuffer is 'foobar', only matches starting with 'foobar' are considered.

You can use different completion styles in different situations, by setting the variable completion-category-overrides. For example, the default setting says to use only basic and substring completion for buffer names.

Options

Case is significant when completing case-sensitive arguments, such as command names. For example, when completing command names, 'AU' does not complete to 'auto-fill-mode'. Case differences are ignored when completing arguments where case does not matter.

When completing file names, case differences are ignored if the variable read-file-name-completion-ignore-case is non-nil. The default value is nil on systems that have case-sensitive file-names, such as GNU Linux; it is non-nil on systems that have case-insensitive file-names, such as Microsoft Windows. When completing buffer names, case differences are ignored if the variable read-buffer-completion-ignore-case is non-nil; the default is nil.

When completing file names, Emacs usually omits certain alternatives that are considered unlikely to be chosen, as determined by the list variable completion-ignored-extensions. Each element in the list should be a string; any file name ending in such a string is ignored as a completion alternative. Any element ending in a slash (/) represents a subdirectory name. The standard value of completion-ignored-extensions has several elements including ".o", ".elc", and "~". For example, if a directory contains 'foo.c' and 'foo.elc', 'foo' completes to 'foo.c'. However, if all possible completions end in "ignored" strings, they are not ignored: in the previous example, 'foo.e' completes to 'foo.elc'. Emacs disregards completion-ignored-extensions when showing completion alternatives in the completion list.

If completion-auto-help is set to nil, the completion commands never display the completion list buffer; you must type ? to display the list. If the value is lazy, Emacs only shows the completion list buffer on the second attempt to complete. In other words, if there is nothing to complete, the first <TAB> echoes 'Next char not unique'; the second <TAB> shows the completion list buffer.

If completion-cycle-threshold is non-nil, completion commands can "cycle" through completion alternatives. Normally, if there is more than one completion alternative for the text in the minibuffer, a completion command completes up to the longest common substring. If you change completion-cycle-threshold to t, the completion command instead completes to the first of those completion alternatives; each subsequent invocation of the completion command replaces that with the next completion alternative, in a cyclic manner. If you give completion-cycle-threshold a numeric value n, completion commands switch to this cycling behavior only when there are n or fewer alternatives.

Icomplete mode presents a constantly-updated display that tells you what completions are available for the text you've entered so far. The command to enable or disable this minor mode is M-x icomplete-mode.

Minibuffer history

Every argument that you enter with the minibuffer is saved in a minibuffer history list so you can easily use it again later. You can use the following arguments to quickly fetch an earlier argument into the minibuffer:

While in the minibuffer, M-p or Up (previous-history-element) moves through the minibuffer history list, one item at a time. Each M-p fetches an earlier item from the history list into the minibuffer, replacing its existing contents. Typing M-n or Down (next-history-element) moves through the minibuffer history list in the opposite direction, fetching later entries into the minibuffer.

If you type M-n in the minibuffer when there are no later entries in the minibuffer history (e.g., if you haven't previously typed M-p), Emacs tries fetching from a list of default arguments: values that you are likely to enter. You can think of this as moving through the "future history" list.

If you edit the text inserted by the M-p or M-n minibuffer history commands, this does not change its entry in the history list. However, the edited argument does go at the end of the history list when you submit it.

You can use M-r (previous-matching-history-element) to search through older elements in the history list, and M-s (next-matching-history-element) to search through newer entries. Each of these commands asks for a regular expression as an argument, and fetches the first matching entry into the minibuffer. See Regexps, for an explanation of regular expressions. A numeric prefix argument n means to fetch the nth matching entry. These commands are unusual, in that they use the minibuffer to read the regular expression argument, even though they are invoked from the minibuffer. An upper-case letter in the regular expression makes the search case-sensitive (see Search Case).

You can also search through the history using an incremental search. See Isearch Minibuffer.

Emacs keeps separate history lists for t kinds of arguments. For example, there is a list for file names, used by all the commands that read file names. Other history lists include buffer names, command names (used by M-x), and command arguments (used by commands like query-replace).

The variable history-length specifies the maximum length of a minibuffer history list; adding a new element deletes the oldest element if the list gets too long. If the value is t, there is no maximum length.

The variable history-delete-duplicates specifies whether to delete duplicates in history. If it's non-nil, adding a new element deletes from the list all other elements that are equal to it. The default is nil.

Repeating minibuffer commands

Every command that uses the minibuffer once is recorded on a special history list, the command history, together with the values of its arguments, so that you can repeat the entire command. In particular, every use of M-x is recorded there, as M-x uses the minibuffer to read the command name.

C-x <ESC> <ESC>

Re-executes a recent minibuffer command from the command history (repeat-complex-command).

M-x list-command-history

Display the entire command history, showing all the commands C-x <ESC> <ESC> can repeat, most recent first.

C-x <ESC> <ESC> re-executes a recent command that used the minibuffer. With no argument, it repeats the last such command. A numeric argument specifies which command to repeat; 1 means the last one, 2 the previous, etc.

C-x <ESC> <ESC> works by turning the previous command into a Lisp expression and then entering a minibuffer initialized with the text for that expression. Even if you don't know Lisp, it will probably be obvious which command is displayed for repetition. If you type only <RET>, that repeats the command unchanged. You can also change the command by editing the Lisp expression before you execute it. The executed command is added to the front of the command history unless it is identical to the most recent item.

Once inside the minibuffer for C-x <ESC> <ESC>, you can use the usual minibuffer history commands (see Minibuffer History) to move through the history list. After finding the desired previous command, you can edit its expression as usual and then execute it by typing <RET>.

Incremental search does not, strictly speaking, use the minibuffer. Therefore, although it behaves like a complex command, it normally does not appear in the history list for C-x <ESC> <ESC>. You can make incremental search commands appear in the history by setting isearch-resume-in-command-history to a non-nil value. See Incremental Search.

The list of previous minibuffer-using commands is stored as a Lisp list in the variable command-history. Each element is a Lisp expression that describes one command and its arguments. Lisp programs can re-execute a command by calling eval with the command-history element.

Entering passwords

Sometimes, you may need to enter a password into Emacs. For instance, when you tell Emacs to visit a file on another machine via a network protocol such as FTP, you often need to supply a password to gain access to the machine (see Remote Files).

Entering a password is similar to using a minibuffer. Emacs displays a prompt in the echo area (such as 'Password: '); after you type the required password, press <RET> to submit it. To prevent others from seeing your password, every character you type is displayed as a dot ('.') instead of its usual form.

Most of the features and commands associated with the minibuffer cannot be used when entering a password. There is no history or completion, and you cannot change windows or perform any other action with Emacs until you have submitted the password.

While you are typing the password, you may press <DEL> to delete backwards, removing the last character entered. C-u deletes everything you have typed so far. C-g quits the password prompt (see Quitting). C-y inserts the current kill into the password (see Killing). You may type either <RET> or <ESC> to submit the password. Any other self-inserting character key inserts the associated character into the password, and all other input is ignored.

Yes or no prompts

An Emacs command may require you to answer a "yes or no" question during the course of its execution. Such queries come in two main varieties.

For the first type of "yes or no" query, the prompt ends with '(y or n)'. Such a query does not actually use the minibuffer; the prompt appears in the echo area, and you answer by typing either 'y' or 'n', which immediately delivers the response. For example, if you type C-x C-w (write-file) to save a buffer, and enter the name of an existing file, Emacs issues a prompt like this:

File `foo.el' exists; overwrite? (y or n)

Because this query does not actually use the minibuffer, the usual minibuffer editing commands cannot be used. However, you can perform some window scrolling operations while the query is active: C-l recenters the selected window; M-v (or PageDown or next) scrolls forward; C-v (or PageUp, or prior) scrolls backward; C-M-v scrolls forward in the next window; and C-M-S-v scrolls backward in the next window. Typing C-g dismisses the query, and quits the command that issued it (see Quitting).

The second type of "yes or no" query is often employed if giving the wrong answer would have serious consequences; it uses the minibuffer, and features a prompt ending with '(yes or no)'. For example, if you invoke C-x k (kill-buffer) on a file-visiting buffer with unsaved changes, Emacs activates the minibuffer with a prompt like this:

Buffer foo.el modified; kill anyway? (yes or no)

To answer, you must type 'yes' or 'no' into the minibuffer, followed by <RET>. The minibuffer behaves as described in the previous sections; you can switch to another window with C-x o, use the history commands M-p and M-f, etc. Type C-g to quit the minibuffer and the querying command.

Running commands by name

Every Emacs command has a name that you can use to run it. For convenience, many commands also have key bindings. You can run those commands by typing the keys, or run them by name. Most Emacs commands have no key bindings, so the only way to run them is by name. (See Key Bindings, for how to set up key bindings.)

By convention, a command name consists of one or more words, separated by hyphens; for example, auto-fill-mode or manual-entry. Command names mostly use complete English words to make them easier to remember.

To run a command by name, start with M-x, type the command name, then terminate it with <RET>. M-x uses the minibuffer to read the command name. The string 'M-x' appears at the beginning of the minibuffer as a prompt to remind you to enter a command name to run. <RET> exits the minibuffer and runs the command. See Minibuffer, for more information on the minibuffer.

You can use completion to enter the command name. For example, to invoke the command forward-char, you can type

M-x forward-char <RET>

or

M-x forw <TAB> c <RET>

Note that forward-char is the same command that you invoke with the key C-f. The existence of a key binding does not stop you from running the command by name.

To cancel the M-x and not run a command, type C-g instead of entering the command name. This takes you back to command level.

To pass a numeric argument to the command you are invoking with M-x, specify the numeric argument before M-x. The argument value appears in the prompt while the command name is being read, and finally M-x passes the argument to that command.

When the command you run with M-x has a key binding, Emacs mentions this in the echo area after running the command. For example, if you type M-x forward-word, the message says that you can run the same command by typing M-f. You can turn off these messages by setting the variable suggest-key-bindings to nil.

In this manual, when we speak of running a command by name, we often omit the <RET> that terminates the name. Thus we might say M-x auto-fill-mode rather than M-x auto-fill-mode <RET>. We mention the <RET> only for emphasis, such as when the command is followed by arguments.

M-x works by running the command execute-extended-command, which is responsible for reading the name of another command and invoking it.

Help

Emacs provides a wide variety of help commands, all accessible through the prefix key C-h (or, equivalently, the function key F1). These help commands are described in the following sections. You can also type C-h C-h to view a list of help commands (help-for-help). You can scroll the list with <SPC> and <DEL>, then type the help command you want. To cancel, type C-g.

Many help commands display their information in a special help buffer. In this buffer, you can type <SPC> and <DEL> to scroll and type <RET> to follow hyperlinks. See Help Mode.

If you are looking for a certain feature, but don't know what it is called or where to look, we recommend three methods. First, try the apropos command, then try searching the manual index, then look in the FAQ and the package keywords.

C-h or F1 means "help" in other contexts as well. For instance, you can type them after a prefix key to view a list of the keys that follow the prefix key. (A few prefix keys don't support C-h in this way, because they define other meanings for it, but they all support F1 for help.)

• Help Summary: Brief list of all Help commands.
• Key Help: Asking what a key does in Emacs.
• Name Help: Asking about a command, variable or function name.
• Apropos: Asking what pertains to a given topic.
• Help Mode: Special features of Help mode and Help buffers.
• Package Keywords: Finding Lisp libraries by keywords (topics).
• Language Help: Help relating to international language support.
• Misc Help: Other help commands.
• Help Files: Commands to display auxiliary help files.
• Help Echo: Help on active text and tooltips ("balloon help").

Help summary

Here is a summary of help commands for accessing the built-in documentation. Most of these are described in more detail in the following sections.

Documentation for a key

The help commands to get information about a key sequence are C-h c (describe-key-briefly) and C-h k (describe-key).

C-h c key displays in the echo area the name of the command that key is bound. For example, C-h c C-f displays 'forward-char'.

C-h k key is similar but gives more information: it displays a help buffer containing the command's documentation string, which describes exactly what the command does.

C-h K key displays the section of the Emacs manual that describes the command corresponding to key.

C-h c, C-h k and C-h K work for any sort of key sequences, including function keys, menus, and mouse events. For instance, after C-h k you can select a menu item from the menu bar, to view the documentation string of the command it runs.

C-h w command <RET> lists the keys that are bound to command. It displays the list in the echo area. If it says the command is not on any key, that means you must use M-x to run it. C-h w runs the command where-is.

Help by command or variable name

C-h f function <RET> (describe-function) displays the documentation of Lisp function function, in a window. Since commands are Lisp functions, you can use this method to view the documentation of any command whose name you know. For example,

C-h f auto-fill-mode <RET>

Displays the documentation of auto-fill-mode. This is the only way to get the documentation of a command that is not bound to any key (one which you would normally run using M-x).

C-h f is also useful for Lisp functions that you use in a Lisp program. For example, if you have only written the expression (make-vector len) and want to check that you are using make-vector properly, type C-h f make-vector <RET>. Because C-h f allows all function names, not only command names, you may find that some of your favorite completion abbreviations that work in M-x don't work in C-h f. An abbreviation that is unique among command names may not be unique among all function names.

If you type C-h f <RET>, it describes the function called by the innermost Lisp expression in the buffer around point, provided that function name is a valid, defined Lisp function. (That name appears as the default while you enter the argument.) For example, if point is located following the text '(make-vector (car x)', the innermost list containing point is the one that starts with '(make-vector', so C-h f <RET> describes the function make-vector. C-h f is also useful only to verify that you spelled a function name correctly. If the minibuffer prompt for C-h f shows the function name from the buffer as the default, it means that name is defined as a Lisp function. Type C-g to cancel the C-h f command if you don't really want to view the documentation.

C-h v (describe-variable) is like C-h f but describes Lisp variables instead of Lisp functions. Its default is the Lisp symbol around or before point, if that is the name of a defined Lisp variable. See Variables.

Help buffers that describe Emacs variables and functions normally have hyperlinks to the corresponding source code, if you have the source files installed (see Hyperlinking).

To find a command's documentation in a manual, use C-h F (Info-goto-emacs-command-node). This knows about various manuals, not only the Emacs manual, and finds the right one.

Apropos

Apropos commands answer questions like, "What are the commands for working with files?" More precisely, you specify an apropos pattern, which means either a word, a list of words, or a regular expression.

Each of the following apropos commands reads an apropos pattern in the minibuffer, searches for items that match the pattern, and displays the results in a different window.

The simplest kind of apropos pattern is one word. Anything containing that word matches the pattern. Thus, to find commands that work on files, type C-h a file <RET>. This displays a list of all command names that contain 'file', including copy-file, find-file, etc. Each command name comes with a brief description and a list of keys you can currently use. In our example, it would say that you can invoke find-file by typing C-x C-f.

For more information about a function definition, variable or symbol property listed in the apropos buffer, you can click it with Mouse-1 or Mouse-2, or move there and type <RET>.

When you specify more than one word in the apropos pattern, a name must contain at least two of the words to match. Thus, if you are looking for commands to kill a chunk of text before point, you could try C-h a kill back backward behind before <RET>. The real command name kill-backward matches that; if there were a command kill-text-before, it would also match, as it contains two of the specified words.

For even greater flexibility, you can specify a regular expression (see Regexps). An apropos pattern is interpreted as a regular expression if it contains any of the regular expression special characters, '^$*+?.\['.

Following the conventions for naming Emacs commands, here are some words that are useful in apropos patterns. Using them in C-h a, you can also get a feel for the naming conventions.

char, line, word, sentence, paragraph, region, page, sexp, list, defun, rect, buffer, frame, window, face, file, dir, register, mode, beginning, end, forward, backward, next, previous, up, down, search, goto, kill, delete, mark, insert, yank, fill, indent, case, change, set, what, list, find, view, describe, default.

If the variable apropos-do-all is non-nil, the apropos commands always behave as if they had been given a prefix argument.

By default, all apropos commands except apropos-documentation list their results in alphabetical order. If the variable apropos-sort-by-scores is non-nil, these commands instead try to guess the relevance of each result, and display the most relevant ones first. The apropos-documentation command lists its results in order of relevance by default; to list them in alphabetical order, change the variable apropos-documentation-sort-by-scores to nil.

Help mode commands

Help buffers provide the same commands as View mode (see View Mode); for instance, <SPC> scrolls forward, and <DEL> scrolls backward. A few special commands are also provided:

When a function name, variable name, or face name (see Faces) appears in the documentation in the help buffer, it is normally an underlined hyperlink. To view the associated documentation, move point there and type <RET> (help-follow), or click the hyperlink with Mouse-1 or Mouse-2. Doing so replaces the contents of the help buffer; to retrace your steps, type C-c C-b (help-go-back).

A help buffer can also contain hyperlinks to Info manuals, source code definitions, and URLs (web pages). The first two are opened in Emacs, and the third using a web browser via the browse-url command (see Browse-URL).

In a help buffer, <TAB> (forward-button) moves point forward to the next hyperlink, while S-<TAB> (backward-button) point back to the previous hyperlink. These commands act cyclically; for instance, typing <TAB> at the last hyperlink moves back to the first hyperlink.

To view all documentation about any symbol in the text, move point to there and type C-c C-c (help-follow-symbol). This shows all available documentation about the symbol: as a variable, function and/or face.

Keyword search for packages

Most optional features in Emacs are grouped into packages. Emacs contains several hundred built-in packages, and more can be installed over the network (see Packages).

To make it easier to find packages related to a topic, most packages are associated with one or more keywords based on what they do. Type C-h p (finder-by-keyword) to bring up a list of package keywords, together with a description of what the keywords mean. To view a list of packages for a given keyword, type <RET> on that line; this displays the list of packages in a Package Menu buffer (see Package Menu).

C-h P (describe-package) prompts for the name of a package, and displays a help buffer describing the attributes of the package and the features that it implements.

Help for international language support

For information on a specific language environment (see Language Environments), type C-h L (describe-language-environment). This displays a help buffer describing the languages supported by the language environment, and listing the associated character sets, coding systems, and input methods, and some sample text for that language environment.

The command C-h h (view-hello-file) displays the file etc/HELLO, which demonstrates various character sets by showing how to say "hello" in many languages.

The command C-h I (describe-input-method) describes an input method—either a specified input method, or by default the input method currently in use. See Input Methods.

The command C-h C (describe-coding-system) describes coding systems: either a specified coding system, or the ones currently in use. See Coding Systems.

Other help commands

C-h i (info) runs the Info program, which browses structured documentation files. The entire Emacs manual is available within Info, along with other manuals for the GNU system. Type h after entering Info to run a tutorial on using Info.

With a numeric argument n, C-h i selects the Info buffer '*info*<n>'. This is useful if you want to browse multiple Info manuals simultaneously. If you specify only C-u as the prefix argument, C-h i prompts for the name of a documentation file, so you can browse a file which doesn't have an entry in the top-level Info menu.

The help commands C-h F function <RET> and C-h K key, described above, enter Info and go straight to the documentation of function or key.

When editing a program, if you have an Info version of the manual for the programming language, you can use C-h S (info-lookup-symbol) to find an entry for a symbol (keyword, function or variable) in the proper manual. The details of how this command works depend on the major mode.

If something surprising happens, and you are not sure what you typed, use C-h l (view-lossage). C-h l displays your last 300 input keystrokes. If you see commands that you don't know, you can use C-h c to find out what they do.

To review recent echo area messages, use C-h e (view-echo-area-messages). This displays the buffer *Messages*, where those messages are kept.

Each Emacs major mode often redefines a few keys and makes other changes in how editing works. C-h m (describe-mode) displays documentation on the current major mode, which normally describes the commands and features that are changed in this mode.

C-h b (describe-bindings) and C-h s (describe-syntax) show other information about the current environment within Emacs. C-h b displays a list of all the key bindings now in effect: first the local bindings of the current minor modes, then the local bindings defined by the current major mode, and finally the global bindings (see Key Bindings). C-h s displays the contents of the syntax table, with explanations of each character's syntax.

You can get a list of subcommands for a particular prefix key by typing C-h (describe-prefix-bindings) after the prefix key. (There are a few prefix keys where this does not work—those that provide their bindings for C-h. One of these is <ESC>, because <ESC> C-h is actually C-M-h, which marks a defun.)

Help files

Apart from the built-in documentation and manuals, Emacs contains other files describing topics like copying conditions, release notes, instructions for debugging and reporting bugs, and so forth. You can use the following commands to view these files. Apart from C-h g, they all have the form C-h C-char.

Help on active text and tooltips

In Emacs, stretches of "active text" (text that does something special in response to mouse clicks or <RET>) often have associated help text. This includes hyperlinks in Emacs buffers, and parts of the mode line. On graphical displays, and some text terminals which support mouse tracking, moving the mouse over the active text displays the help text as a tooltip. See Tooltips.

On terminals that don't support mouse-tracking, you can display the help text for active buffer text at point by typing C-h . (display-local-help). This shows the help text in the echo area. To display help text automatically whenever it is available at point, set the variable help-at-pt-display-when-idle to t.

The mark and the region

Many Emacs commands operate on an arbitrary contiguous part of the current buffer. To specify the text for such a command to operate on, you set the mark at one end of it, and move point to the other end. The text between point and the mark is called the region. The region always extends between point and the mark, no matter which one comes earlier in the text; each time you move point, the region changes.

Setting the mark at a position in the text also activates it. When the mark is active, we say also that the region is active; Emacs indicates its extent by highlighting the text within it, using the region face (see Face Customization).

After certain non-motion commands, including any command that changes the text in the buffer, Emacs automatically deactivates the mark; this turns off the highlighting. You can also explicitly deactivate the mark at any time, by typing C-g (see Quitting).

The above default behavior is known as Transient Mark mode. Disabling Transient Mark mode switches Emacs to an alternative behavior, where the region is usually not highlighted. See Disabled Transient Mark.

Setting the mark in one buffer has no effect on the marks in other buffers. When you return to a buffer with an active mark, the mark is at the same place as before. When multiple windows show the same buffer, they can have different values of point, and thus different regions, but they all share one common mark position. See Windows. Ordinarily, only the selected window highlights its region; however, if the variable highlight-nonselected-windows is non-nil, each window highlights its own region.

• Setting Mark: Commands to set the mark.
• Marking Objects: Commands to put region around textual units.
• Using Region: Summary of ways to operate on contents of the region.
• Mark Ring: Previous mark positions saved so you can go back there.
• Global Mark Ring: Previous mark positions in various buffers.
• Shift Selection: Using shifted cursor motion keys.
• Disabled Transient Mark: Leaving regions unhighlighted by default.

Setting up the mark

Here are some commands for setting the mark:

The most common way to set the mark is with C-<SPC> (set-mark-command). This sets the mark where point is, and activates it. You can then move point away, leaving the mark behind.

For example, suppose you want to convert part of the buffer to uppercase. To accomplish this, go to one end of the desired text, type C-<SPC>, and move point until the desired portion of text is highlighted. Now type C-x C-u (upcase-region). This converts the text in the region to uppercase, and then deactivates the mark.

Whenever the mark is active, you can deactivate it by typing C-g (see Quitting). Most commands that operate on the region also automatically deactivate the mark, like C-x C-u in the above example.

Instead of setting the mark to operate on a region, you can also use it to "remember" a position in the buffer (by typing C-<SPC> C-<SPC>), and later jump back there (by typing C-u C-<SPC>). See Mark Ring, for details.

The command C-x C-x (exchange-point-and-mark) exchanges the positions of point and the mark. C-x C-x is useful when you are satisfied with the position of point but want to move the other end of the region (where the mark is). Using C-x C-x a second time, if necessary, puts the mark at the new position with point back at its original position. Normally, if the mark is inactive, this command first reactivates the mark wherever it was last set, to ensure that the region is left highlighted. However, if you call it with a prefix argument, it leaves the mark inactive and the region unhighlighted; you can use this to jump to the mark in a manner similar to C-u C-<SPC>. You can also set the mark with the mouse. If you press the left mouse button (down-mouse-1) and drag the mouse across a range of text, this sets the mark where you first pressed the mouse button and puts point where you release it. Alternatively, clicking the right mouse button (mouse-3) sets the mark at point and then moves point to where you clicked. See Mouse Commands, for a more detailed description of these mouse commands.

Finally, you can set the mark by holding down the Shift key while typing certain cursor motion commands (such as S-right, S-C-f, S-C-n, etc.). This is called shift-selection. It sets the mark at point before moving point, but only if there is no active mark set via shift-selection. The mark set by mouse commands and by shift-selection behaves slightly differently from the usual mark: any subsequent unshifted cursor motion command deactivates it automatically. For details, See Shift Selection.

Many commands that insert text, such as C-y (yank), set the mark at the other end of the inserted text, without activating it. This lets you easily return to that position (see Mark Ring). You can tell that a command does this when it shows 'Mark set' in the echo area.

Under X, every time the active region changes, Emacs saves the text in the region to the primary selection. This lets you insert that text into other X applications with mouse-2 clicks. See Primary Selection.

Commands to mark textual objects

Here are commands for placing point and the mark around a textual object such as a word, list, paragraph or page:

M-@ (mark-word) sets the mark at the end of the next word (see Words, for information about words). Repeated invocations of this command extend the region by advancing the mark one word at a time. As an exception, if the mark is active and located before point, M-@ moves the mark backwards from its current position one word at a time.

This command also accepts a numeric argument n, which tells it to advance the mark by n words. A negative argument moves the mark back by n words.

Similarly, C-M-@ (mark-sexp) puts the mark at the end of the next balanced expression (see Expressions). Repeated invocations extend the region to subsequent expressions, while positive or negative numeric arguments move the mark forward or backward by the specified number of expressions.

The other commands in the above list set both point and mark, so as to delimit an object in the buffer. M-h (mark-paragraph) marks paragraphs (see Paragraphs), C-M-h (mark-defun) marks top-level definitions (see Moving by Defuns), and C-x C-p (mark-page) marks pages (see Pages). Repeated invocations again play the same role, extending the region to consecutive objects; similarly, numeric arguments specify how many objects to move the mark by.

C-x h (mark-whole-buffer) sets up the entire buffer as the region, by putting point at the beginning and the mark at the end.

Operating on the region

Once you have a region, here are some of the ways you can operate on it:

• Kill it with C-w (see Killing).
• Copy it to the kill ring with M-w (see Yanking).
• Convert case with C-x C-l or C-x C-u (see Case).
• Undo changes within it using C-u C-/ (see Undo).
• Replace text within it using M-% (see Query Replace).
• Indent it with C-x <TAB> or C-M-\ (see Indentation).
• Fill it as text with M-x fill-region (see Filling).
• Check the spelling of words within it with M-$ (see Spelling).
• Evaluate it as Lisp code with M-x eval-region (see Lisp Eval).
• Save it in a register with C-x r s (see Registers).
• Save it in a buffer or a file (see Accumulating Text).

Some commands have a default behavior when the mark is inactive, but operate on the region if the mark is active. For example, M-$ (ispell-word) normally checks the spelling of the word at point, but it checks the text in the region if the mark is active (see Spelling). Normally, such commands use their default behavior if the region is empty (i.e., if mark and point are at the same position). If you want them to operate on the empty region, change the variable use-empty-active-region to t.

As described in Erasing, the <DEL> (backward-delete-char) and delete (delete-forward-char) commands also act this way. If the mark is active, they delete the text in the region. (As an exception, if you supply a numeric argument n, where n is not one, these commands delete n characters regardless of whether the mark is active). If you change the variable delete-active-region to nil, then these commands don't act differently when the mark is active. If you change the value to kill, these commands kill the region instead of deleting it (see Killing).

Other commands always operate on the region, and have no default behavior. Such commands usually have the word region in their names, like C-w (kill-region) and C-x C-u (upcase-region). If the mark is inactive, they operate on the "inactive region"—that is, on the text between point and the position at which the mark was last set (see Mark Ring). To disable this behavior, change the variable mark-even-if-inactive to nil. Then these commands will instead signal an error if the mark is inactive.

By default, text insertion occurs normally even if the mark is active; for example, typing a inserts the character 'a', then deactivates the mark. If you enable Delete Selection mode, a minor mode, then inserting text while the mark is active causes the text in the region to be deleted first. To toggle Delete Selection mode on or off, type M-x delete-selection-mode.

The mark ring

Each buffer remembers previous locations of the mark, in the mark ring. Commands that set the mark also push the old mark to this ring. One of the uses of the mark ring is to remember spots where you may want to go back.

C-<SPC> C-<SPC>

Set the mark, pushing it onto the mark ring, without activating it.

C-u C-<SPC>

Move point to where the mark was, and restore the mark from the ring of former marks.

The command C-<SPC> C-<SPC> is handy when you want to use the mark to remember a position where you may want to return. It pushes the current point onto the mark ring, without activating the mark (which would cause Emacs to highlight the region). This is actually two consecutive invocations of C-<SPC> (set-mark-command); the first C-<SPC> sets the mark, and the second C-<SPC> deactivates it. (When Transient Mark mode is off, C-<SPC> C-<SPC> instead activates Transient Mark mode temporarily; see Disabled Transient Mark.)

To return to a marked position, use set-mark-command with a prefix argument: C-u C-<SPC>. This moves point to where the mark was, and deactivates the mark if it was active. Each subsequent C-u C-<SPC> jumps to a prior position stored in the mark ring. The positions you move through in this way are not lost; they go to the end of the ring.

If you set set-mark-command-repeat-pop to non-nil, then immediately after you type C-u C-<SPC>, you can type C-<SPC> instead of C-u C-<SPC> to cycle through the mark ring. By default, set-mark-command-repeat-pop is nil.

Each buffer has its own mark ring. All editing commands use the current buffer's mark ring. In particular, C-u C-<SPC> always stays in the same buffer.

The variable mark-ring-max specifies the maximum number of entries to keep in the mark ring. This defaults to 16 entries. If that many entries exist and another one is pushed, the earliest one in the list is discarded. Repeating C-u C-<SPC> cycles through the positions currently in the ring.

To move back to the same place over and over, the mark ring may not be convenient enough. If so, you can record the position in a register for later retrieval (see Saving Positions in Registers).

The global mark ring

In addition to the ordinary mark ring that belongs to each buffer, Emacs has a single global mark ring. Each time you set a mark, this is recorded in the global mark ring in addition to the current buffer's own mark ring, if you have switched buffers since the previous mark setting. Hence, the global mark ring records a sequence of buffers that you are in, and, for each buffer, a place where you set the mark. The length of the global mark ring is controlled by global-mark-ring-max, and is 16 by default.

The command C-x C-<SPC> (pop-global-mark) jumps to the buffer and position of the latest entry in the global ring. It also rotates the ring, so that successive uses of C-x C-<SPC> take you to earlier buffers and mark positions.

Shift selection

If you hold down the Shift key while typing a cursor motion command, this sets the mark before moving point, so that the region extends from the original position of point to its new position. This feature is referred to as shift-selection. It is similar to the way text is selected in other editors.

The mark set via shift-selection behaves a little differently from what we have described above. Firstly, in addition to the usual ways of deactivating the mark (such as changing the buffer text or typing C-g), the mark is deactivated by any unshifted cursor motion command. Secondly, any subsequent shifted cursor motion command avoids setting the mark anew. Therefore, a series of shifted cursor motion commands will continuously adonly the region.

Shift-selection only works if the shifted cursor motion key is not already bound to a separate command (see Customization). For example, if you bind S-C-f to another command, typing S-C-f runs that command instead of performing a shift-selected version of C-f (forward-char).

A mark set via mouse commands behaves the same as a mark set via shift-selection (see Setting Mark). For example, if you specify a region by dragging the mouse, you can continue to extend the region using shifted cursor motion commands. In either case, any unshifted cursor motion command deactivates the mark.

To turn off shift-selection, set shift-select-mode to nil. Doing so does not disable setting the mark via mouse commands.

Disabling transient mark mode

The default behavior of the mark and region, where setting the mark activates it and highlights the region, is called Transient Mark mode. This is a minor mode that is enabled by default. It can be toggled with M-x transient-mark-mode, or with the 'Active Region Highlighting' menu item in the 'Options' menu. Turning it off switches Emacs to an alternative mode of operation:

• Setting the mark, with commands like C-<SPC> or C-x C-x, does not highlight the region. Therefore, you can't tell by looking where the mark is located; you have to remember.
• The usual solution to this problem is to set the mark and then use it soon, before you forget where it is. You can also check where the mark uses C-x C-x, which exchanges the positions of the point and the mark (see Setting Mark).
• Some commands, which ordinarily act on the region when the mark is active, no longer do so. For example, normally M-% (query-replace) performs replacements in the region, if the mark is active. When Transient Mark mode is off, it always operates from point to the end of the buffer. Commands that act this way are identified in their documentation.

While Transient Mark mode is off, you can activate it temporarily using C-<SPC> C-<SPC> or C-u C-x C-x.

C-<SPC> C-<SPC>

Set the mark at point (like plain C-<SPC>) and enable Transient Mark mode only once, until the mark is deactivated. (This is not a separate command; you are using the C-<SPC> command twice.)

C-u C-x C-x

Exchange point and mark, activate the mark and enable Transient Mark mode temporarily, until the mark is next deactivated. (This is the C-x C-x command, exchange-point-and-mark, with a prefix argument.)

These commands set or activate the mark, and enable Transient Mark mode only until the mark is deactivated. One reason you may want to use them is that some commands operate on the entire buffer instead of the region when Transient Mark mode is off. Enabling Transient Mark mode momentarily gives you a way to use these commands on the region.

When you specify a region with the mouse (see Setting Mark), or with shift-selection (see Shift Selection), this likewise activates Transient Mark mode temporarily and highlights the region.

Killing and moving text

In Emacs, killing means erasing text and copying it into the kill ring. Yanking means bringing text from the kill ring back into the buffer. (Some applications use the terms "cutting" and "pasting" for similar operations.) The kill ring is so-named because it can be visualized as a set of blocks of text arranged in a ring, which you can access in cyclic order. See Kill Ring.

Killing and yanking are the most common way to move or copy text within Emacs. It is very versatile, because there are commands for killing many different types of syntactic units.

• Deletion and Killing: Commands that remove text.
• Yanking: Commands that insert text.
• Cut and Paste: Clipboard and selections on graphical displays.
• Accumulating Text: Other methods to add text to the buffer.
• Rectangles: Operating on text in rectangular areas.
• CUA Bindings: Using C-x/C-c/C-v to kill and yank.

Deletion and killing

Most commands which erase text from the buffer save it in the kill ring. These are known as kill commands, and their names normally contain the word 'kill' (e.g., kill-line). The kill ring stores several recent kills, not only the last one, so killing is a very safe operation: you don't have to worry much about losing text that you previously killed. The kill ring is shared by all buffers, so text that is killed in one buffer can be yanked into another buffer.

When you use C-/ (undo) to undo a kill command (see Undo), that brings the killed text back into the buffer, but does not remove it from the kill ring.

On graphical displays, killing text also copies it to the system clipboard. See Cut and Paste.

Commands that erase text but do not save it in the kill ring are known as delete commands; their names usually contain the word 'delete'. These include C-d (delete-char) and <DEL> (delete-backward-char), which delete only one character at a time, and those commands that delete only spaces or newlines. Commands that erase significant amounts of nontrivial data generally do a kill operation instead.

You can also use the mouse to kill and yank. See Cut and Paste.

• Deletion: Commands for deleting small amounts of text and blank areas.
• Killing by Lines: How to kill entire lines of text at one time.
• Other Kill Commands: Commands to kill large regions of text and syntactic units such as words and sentences.
• Options: Options that affect killing.

Deletion

Deletion means erasing text and not saving it in the kill ring. For the most part, the Emacs commands that delete text are those that erase only one character or only whitespace.

We have already described the basic deletion commands <DEL> (delete-backward-char), delete (delete-forward-char), and C-d (delete-char). See Erasing. With a numeric argument, they delete the specified number of characters. If the numeric argument is omitted or one, they delete all the text in the region if it's active (see Using Region).

The other delete commands are those that delete only whitespace characters: spaces, tabs and newlines. M-\ (delete-horizontal-space) deletes all the spaces and tab characters before and after point. With a prefix argument, this only deletes spaces and tab characters before point. M-<SPC> (only-one-space) does likewise but leaves a single space before point, regardless of the number of spaces that existed previously (even if there were none before). With a numeric argument n, it leaves n spaces before point if n is positive; if n is negative, it deletes newlines in addition to spaces and tabs, leaving -n spaces before point.

C-x C-o (delete-blank-lines) deletes all blank lines after the current line. If the current line is blank, it deletes all blank lines preceding the current line as well (leaving one blank line, the current line). On a solitary blank line, it deletes that line.

M-^ (delete-indentation) joins the current line and the previous line, by deleting a newline and all surrounding spaces, usually leaving a single space. See M-^.

Killing by lines

C-k

Kill rest of line or one or more lines (kill-line).

C-S-backspace

Kill an entire line at once (kill-whole-line)

The simplest kill command is C-k (kill-line). If used at the end of a line, it kills the line-ending newline character, merging the next line into the current one (thus, a blank line is entirely removed). Otherwise, C-k kills all the text from point up to the end of the line; if point was originally at the beginning of the line, this leaves the line blank.

Spaces and tabs at the end of the line are ignored when deciding which case applies. As long as point is after the last visible character in the line, you can be sure that C-k kills the newline. To kill an entire non-blank line, go to the beginning and type C-k twice.

In this context, "line" means a logical text line, not a screen line (see Continuation Lines).

When C-k is given a positive argument n, it kills n lines and the newlines that follow them (text on the current line before point is not killed). With a negative argument −n, it kills n lines preceding the current line, together with the text on the current line before point. C-k with an argument of zero kills the text before point on the current line.

If the variable kill-whole-line is non-nil, C-k at the very beginning of a line kills the entire line including the following newline. This variable is normally nil.

C-S-backspace (kill-whole-line) kills a whole line including its newline, regardless of the position of point in the line. Note that many text terminals prevent you from typing the key sequence C-S-backspace.

Other kill commands

One of the commonly-used kill commands is C-w (kill-region), which kills the text in the region (see Mark). Similarly, M-w (kill-ring-save) copies the text in the region into the kill ring without removing it from the buffer. If the mark is inactive when you type C-w or M-w, the command acts on the text between point and where you last set the mark (see Using Region).

Emacs also provides commands to kill specific syntactic units: words, with M-<DEL> and M-d (see Words); balanced expressions, with C-M-k (see Expressions); and sentences, with C-x <DEL> and M-k (see Sentences).

The command M-z (zap-to-char) combines killing with searching: it reads a character and kills from point up to (and including) the next occurrence of that character in the buffer. A numeric argument acts as a repeat count; a negative argument means to search backward and kill text before point.

Options for killing

Some specialized buffers contain read-only text, which cannot be modified and therefore cannot be killed. The kill commands work specially in a read-only buffer: they move over text and copy it to the kill ring, without actually deleting it from the buffer. Normally, they also beep and display an error message when this happens. But if you set the variable kill-read-only-ok to a non-nil value, they only print a message in the echo area to explain why the text has not been erased.

If you change the variable kill-do-not-save-duplicates to a non-nil value, identical subsequent kills yield a single kill-ring entry, without duplication.

Yanking

Yanking means reinserting text previously killed. The usual way to move or copy text is to kill it and then yank it elsewhere.

The basic yanking command is C-y (yank). It inserts the most recent kill, leaving the cursor at the end of the inserted text. It also sets the mark at the beginning of the inserted text, without activating the mark; this lets you jump easily to that position, if you want, with C-u C-<SPC> (see Mark Ring).

With a plain prefix argument (C-u C-y), the command instead leaves the cursor in front of the inserted text, and sets the mark at the end. Using any other prefix argument specifies an earlier kill; e.g., C-u 4 C-y reinserts the fourth most recent kill. See Earlier Kills.

On graphical displays, C-y first checks if another application has placed any text in the system clipboard more recently than the last Emacs kill. If so, it inserts the clipboard's text instead. Thus, Emacs effectively treats "cut" or "copy" clipboard operations performed in other applications like Emacs kills, except that they are not recorded in the kill ring. See Cut and Paste, for details.

• Kill Ring: Where killed text is stored.
• Earlier Kills: Yanking something killed some time ago.
• Appending Kills: Several kills in a row all yank together.

The kill ring

The kill ring is a list of blocks of text that were previously killed. There is only one kill ring, shared by all buffers, so you can kill text in one buffer and yank it in another buffer. This is the usual way to move text from one buffer to another. (There are other methods: for instance, you could store the text in a register; see Registers. See Accumulating Text, for some other ways to move text around.)

The maximum number of entries in the kill ring is controlled by the variable kill-ring-max. The default is 60. If you make a new kill when this limit is reached, Emacs makes room by deleting the oldest entry in the kill ring.

The actual contents of the kill ring are stored in a variable named kill-ring; you can view the entire contents of the kill ring with C-h v kill-ring.

Yanking earlier kills

As explained in Yanking, you can use a numeric argument to C-y to yank text that is no longer the most recent kill. This is useful if you remember which kill ring entry you want. If you don't, you can use the M-y (yank-pop) command to cycle through the possibilities.

If the previous command was a yank command, M-y takes the text that was yanked and replaces it with the text from an earlier kill. So, to recover the text of the next-to-the-last kill, first use C-y to yank the last kill, and then use M-y to replace it with the previous kill. M-y is allowed only after a C-y or another M-y.

You can understand M-y in terms of a "last yank" pointer which points at an entry in the kill ring. Each time you kill, the "last yank" pointer moves to the newly made entry at the front of the ring. C-y yanks the entry which the "last yank" pointer points to. M-y moves the "last yank" pointer to a different entry, and the text in the buffer changes to match. Enough M-y commands can move the pointer to any entry in the ring, so you can get any entry into the buffer. Eventually the pointer reaches the end of the ring; the next M-y loops back around to the first entry again.

M-y moves the "last yank" pointer around the ring, but it does not change the order of the entries in the ring, which always runs from the most recent kill at the front to the oldest one still remembered.

M-y can take a numeric argument, which tells it how many entries to advance the "last yank" pointer by. A negative argument moves the pointer toward the front of the ring; from the front of the ring, it moves "around" to the last entry and continues forward from there.

Once the text you are looking for is brought into the buffer, you can stop doing M-y commands and it stays there. It's only a copy of the kill ring entry, so editing it in the buffer does not change what's in the ring. As long as no new killing is done, the "last yank" pointer remains at the same place in the kill ring, so repeating C-y will yank another copy of the same previous kill.

When you call C-y with a numeric argument, that also sets the "last yank" pointer to the entry that it yanks.

Appending kills

Normally, each kill command pushes a new entry onto the kill ring. However, two or more kill commands in a row combine their text into a single entry, so that a single C-y yanks all the text as a unit, only as it was before it was killed.

Thus, if you want to yank text as a unit, you need not kill all of it with one command; you can keep killing line after line, or word after word, until you have killed it all, and you can still get it all back at once.

Commands that kill forward from point add onto the end of the previous killed text. Commands that kill backward from point add text onto the beginning. This way, any sequence of mixed forward and backward kill commands puts all the killed text into one entry without rearrangement. Numeric arguments do not break the sequence of appending kills. For example, suppose the buffer contains this text:

This is a line -!-of sample text.

With point shown by -!-. If you type M-d M-<DEL> M-d M-<DEL>, killing alternately forward and backward, you end up with 'a line of sample' as one entry in the kill ring, and 'This is text' in the buffer. (Note the double space between 'is' and 'text', which you can clean up with M-<SPC> or M-q.)

Another way to kill the same text is to move back two words with M-b M-b, then kill all four words forward with C-u M-d. This produces the same results in the buffer and in the kill ring. M-f M-f C-u M-<DEL> kills the same text, all going backward; once again, the result is the same. The text in the kill ring entry always has the same order that it had in the buffer before you killed it.

If a kill command is separated from the last kill command by other commands (not only numeric arguments), it starts a new entry on the kill ring. But you can force it to append by first typing the command C-M-w (append-next-kill) right before it. The C-M-w tells the following command, if it's a kill command, to append the text it kills to the last killed text, instead of starting a new entry. With C-M-w, you can kill several separated pieces of text and accumulate them to be yanked back in one place.

A kill command following M-w (kill-ring-save) does not append to the text that M-w copied into the kill ring.

Cut and paste operations on graphical displays

In most graphical desktop environments, you can transfer data (usually text) between different applications using a system facility called the clipboard. On X, two other similar facilities are available: the primary selection and the secondary selection. When Emacs is run on a graphical display, its kill and yank commands integrate with these facilities, so that you can easily transfer text between Emacs and other graphical applications.

By default, Emacs uses UTF-8 as the coding system for inter-program text transfers. If you find that the pasted text is not what you expected, you can specify another coding system by typing C-x <RET> x or C-x <RET> X. You can also request a different data type by customizing x-select-request-type. See Communication Coding.

• Clipboard: How Emacs uses the system clipboard.
• Primary Selection: The temporarily selected text selection.
• Secondary Selection: Cutting without altering point and mark.

Using The clipboard

The clipboard is the facility that most graphical applications use for "cutting and pasting". When the clipboard exists, the kill and yank commands in Emacs make use of it.

When you kill some text with a command such as C-w (kill-region), or copy it to the kill ring with a command such as M-w (kill-ring-save), that text is also put in the clipboard.

When an Emacs kill command puts text in the clipboard, the existing clipboard contents are normally lost. Optionally, you can change save-interprogram-paste-before-kill to t. Then Emacs will first save the clipboard to its kill ring, preventing you from losing the old clipboard data, at the risk of high memory consumption if that data turns out to be large.

Yank commands, such as C-y (yank), also use the clipboard. If another application "owns" the clipboard, i.e., if you cut or copied text there more recently than your last kill command in Emacs, Emacs yanks from the clipboard instead of the kill ring.

Normally, rotating the kill ring with M-y (yank-pop) does not alter the clipboard. However, if you change yank-pop-change-selection to t, then M-y saves the new yank to the clipboard.

To prevent kill and yank commands from accessing the clipboard, change the variable x-select-enable-clipboard to nil.

Many X desktop environments support a feature called the clipboard manager. If you exit Emacs while it is the current "owner" of the clipboard data, and there is a clipboard manager running, Emacs transfers the clipboard data to the clipboard manager so that it is not lost. In some circumstances, this may cause a delay when exiting Emacs; if you want to prevent Emacs from transferring data to the clipboard manager, change the variable x-select-enable-clipboard-manager to nil.

Before Emacs 24, the kill and yank commands used the primary selection (see Primary Selection), not the clipboard. If you prefer this behavior, change x-select-enable-clipboard to nil, x-select-enable-primary to t, and mouse-drag-copy-region to t. In this case, you can use the following commands to act explicitly on the clipboard: clipboard-kill-region kills the region and saves it to the clipboard; clipboard-kill-ring-save copies the region to the kill ring and saves it to the clipboard; and clipboard-yank yanks the contents of the clipboard at point.

Cut and paste with other window applications

Under the X Window System, there exists a primary selection containing the last stretch of text selected in an X application (usually by dragging the mouse). Typically, this text can be inserted into other X applications by mouse-2 clicks. The primary selection is separate from the clipboard. Its contents are more "fragile"; they are overwritten each time you select text with the mouse, whereas the clipboard is only overwritten by explicit "cut" or "copy" commands.

Under X, whenever the region is active (see Mark), the text in the region is saved in the primary selection. This applies regardless of whether the region was made by dragging or clicking the mouse (see Mouse Commands), or by keyboard commands (e.g., by typing C-<SPC> and moving point; see Setting Mark).

If you change the variable select-active-regions to only, Emacs saves only temporarily active regions to the primary selection, i.e., those made with the mouse or with shift selection (see Shift Selection). If you change select-active-regions to nil, Emacs avoids saving active regions to the primary selection entirely.

To insert the primary selection into an Emacs buffer, click mouse-2 (mouse-yank-primary) where you want to insert it. See Mouse Commands.

Microsoft Windows provides no primary selection, but Emacs emulates it within a single Emacs session by storing the selected text internally. Therefore, all the features and commands related to the primary selection work on Windows as they do on X, for cutting and pasting in the same session, but not across Emacs sessions or with other applications.

Secondary selection

In addition to the primary selection, the X Window System provides a second similar facility known as the secondary selection. Nowadays, few X applications make use of the secondary selection, but you can access it using the following Emacs commands:

Double or triple clicking of M-Mouse-1 operates on words and lines, much like Mouse-1.

If mouse-yank-at-point is non-nil, M-Mouse-2 yanks at point. Then it does not matter precisely where you click, or even which of the frame's windows you click. See Mouse Commands.

Accumulating text

Usually we copy or move text by killing it and yanking it, but there are other convenient methods for copying one block of text in many places, or for copying many scattered blocks of text into one place. Here we describe the commands to accumulate scattered pieces of text into a buffer or into a file.

To accumulate text into a buffer, use M-x append-to-buffer. This reads a buffer name, then inserts a copy of the region into the buffer specified. If you specify a nonexistent buffer, append-to-buffer creates the buffer. The text is inserted wherever point is in that buffer. If you are using the buffer for editing, the copied text goes into the middle of the text of the buffer, starting from wherever point happens to be at that moment.

Point in that buffer is left at the end of the copied text, so successive uses of append-to-buffer accumulate the text in the specified buffer in the same order as they were copied. Strictly speaking, append-to-buffer does not always append to the text already in the buffer—it appends only if point in that buffer is at the end. However, if append-to-buffer is the only command you use to alter a buffer, then point is always at the end.

M-x prepend-to-buffer is only like append-to-buffer except that point in the other buffer is left before the copied text, so successive prependings add text in reverse order. M-x copy-to-buffer is similar, except that any existing text in the other buffer is deleted, so the buffer is left containing only the text newly copied into it.

The command M-x insert-buffer can retrieve the accumulated text from another buffer. This prompts for the name of a buffer, and inserts a copy of all the text in that buffer into the current buffer at point, leaving point at the beginning of the inserted text. It also adds the position of the end of the inserted text to the mark ring, without activating the mark. See Buffers, for background information on buffers.

Instead of accumulating text in a buffer, you can append text directly into a file with M-x append-to-file. This prompts for a file name, and adds the text of the region to the end of the specified file. The file is changed immediately on disk.

Use append-to-file only with files that are not being visited in Emacs. Using it on a file that you are editing in Emacs would change the file behind Emacs's back, which can lead to losing some of your editing.

Another way to move text around is to store it in a register. See Registers.

Rectangles

Rectangle commands operate on rectangular areas of the text: all the characters between a certain pair of columns, in a certain range of lines. Emacs has commands to kill rectangles, yank killed rectangles, clear them out, fill them with blanks or text, or delete them. Rectangle commands are useful with text in multicolumn formats, and for changing text into or out of such formats.

To specify a rectangle for a command to work on, set the mark at one corner and point at the opposite corner. The rectangle thus specified is called the region-rectangle. If point and the mark are in the same column, the region-rectangle is empty. If they are in the same line, the region-rectangle is one line high.

The region-rectangle is controlled in much the same way as the region is controlled. But remember that a given combination of point and mark values can be interpreted either as a region or as a rectangle, depending on the command that uses them.

The rectangle operations fall into two classes: commands to erase or insert rectangles, and commands to make blank rectangles.

There are two ways to erase the text in a rectangle: C-x r d (delete-rectangle) to delete the text outright, or C-x r k (kill-rectangle) to remove the text and save it as the last killed rectangle. In both cases, erasing the region-rectangle is like erasing the specified text on each line of the rectangle; if there is any following text on the line, it moves backwards to fill the gap.

"Killing" a rectangle is not killing in the usual sense; the rectangle is not stored in the kill ring, but in a special place that only records the most recent rectangle killed. This is because yanking a rectangle is so different from yanking linear text that different yank commands have to be used. Yank-popping is not defined for rectangles.

C-x r M-w (copy-rectangle-as-kill) is the equivalent of M-w for rectangles: it records the rectangle as the "last killed rectangle", without deleting the text from the buffer.

To yank the last killed rectangle, type C-x r y (yank-rectangle). The rectangle's first line is inserted at point, the rectangle's second line is inserted at the same horizontal position one line vertically below, etc. The number of lines affected is determined by the height of the saved rectangle.

For example, you can convert two single-column lists into a double-column list by killing one of the single-column lists as a rectangle, and then yanking it beside the other list.

You can also copy rectangles into and out of registers with C-x r r r and C-x r i r. See Rectangle Registers.

There are two commands you can use for making blank rectangles: C-x r c (clear-rectangle) blanks out existing text in the region-rectangle, and C-x r o (open-rectangle) inserts a blank rectangle.

M-x delete-whitespace-rectangle deletes horizontal whitespace starting from a particular column. This applies to each of the lines in the rectangle, and the column is specified by the left edge of the rectangle. The right edge of the rectangle does not make any difference to this command.

The command C-x r N (rectangle-number-lines) inserts line numbers along the left edge of the region-rectangle. Normally, the numbering begins from 1 (for the first line of the rectangle). With a prefix argument, the command prompts for a number to begin from, and for a format string with which to print the numbers.

The command C-x r t (string-rectangle) replaces the contents of a region-rectangle with a string on each line. The string's width need not be the same as the width of the rectangle. If the string's width is less, the text after the rectangle shifts left; if the string is wider than the rectangle, the text after the rectangle shifts right.

The command M-x string-insert-rectangle is similar to string-rectangle, but inserts the string on each line, shifting the original text to the right.

CUA bindings

The command M-x cua-mode sets up key bindings that are compatible with the Common User Access (CUA) system used in other applications.

When CUA mode is enabled, the keys C-x, C-c, C-v, and C-z invoke commands that cut (kill), copy, paste (yank), and undo respectively. The C-x and C-c keys perform cut and copy only if the region is active. Otherwise, they still act as prefix keys, so that standard Emacs commands like C-x C-c still work. Note that this means the variable mark-even-if-inactive has no effect for C-x and C-c (see Using Region).

To enter an Emacs command like C-x C-f while the mark is active, use one of the following methods: either hold Shift together with the prefix key, e.g., S-C-x C-f, or quickly type the prefix key twice, e.g., C-x C-x C-f.

To disable the overriding of standard Emacs binding by CUA mode, while retaining the other features of CUA mode described below, set the variable cua-enable-cua-keys to nil.

In CUA mode, typed text replaces the active region as in Delete-Selection mode (see Mouse Commands).

CUA mode provides enhanced rectangle support with visible rectangle highlighting. Use C-<RET> to start a rectangle, extend it using the movement commands, and cut or copy it using C-x or C-c. <RET> moves the cursor to the next (clockwise) corner of the rectangle, so you can easily expand it in any direction. Normal text you type is inserted to the left or right of each line in the rectangle (on the same side as the cursor).

With CUA you can easily copy text and rectangles into and out of registers by providing a one-digit numeric prefix to the kill, copy, and yank commands, e.g., C-1 C-c copies the region into register 1, and C-2 C-v yanks the contents of register 2.

CUA mode also has a global mark feature that allows easy moving and copying of text between buffers. Use C-S-<SPC> to toggle the global mark on and off. When the global mark is on, all text that you kill or copy is automatically inserted at the global mark, and text you type is inserted at the global mark rather than at the current position.

For example, to copy words from various buffers into a word list in a given buffer, set the global mark in the target buffer, then navigate to each of the words you want in the list, mark it (e.g., with S-M-f), copy it to the list with C-c or M-w, and insert a newline after the word in the target list by pressing <RET>.

Registers

Emacs registers are compartments where you can save text, rectangles, positions, and other things for later use. Once you save text or a rectangle in a register, you can copy it into the buffer once, or many times; once you save a position in a register, you can jump back to that position once, or many times.

Each register has a name that consists of a single character, which we will denote by r; r is a letter (such as 'a') or a number (such as '1'); case matters, so register 'a' is not the same as register 'A'.

A register can store a position, a piece of text, a rectangle, a number, a window configuration, or a file name, but only one thing at any given time. Whatever you store in a register remains there until you store something else in that register. To see what register r contains, use M-x view-register:

Bookmarks record files and positions in them, so you can return to those positions when you look at the file again. Bookmarks are similar in spirit to registers, so they are also documented in this chapter.

• Position Registers: Saving positions in registers.
• Text Registers: Saving text in registers.
• Rectangle Registers: Saving rectangles in registers.
• Configuration Registers: Saving window configurations in registers.
• Number Registers: Numbers in registers.
• File Registers: File names in registers.
• Bookmarks: Bookmarks are like registers, but persistent.

Saving positions in registers

Typing C-x r <SPC> (point-to-register), followed by a character r, saves both the position of point and the current buffer in register r. The register retains this information until you store something else in it.

The command C-x r j r switches to the buffer recorded in register r, and moves point to the recorded position. The contents of the register are not changed, so you can jump to the saved position any number of times.

If you use C-x r j to go to a saved position, but the buffer it was saved from is killed, C-x r j tries to create the buffer again by visiting the same file. Of course, this works only for buffers that were visiting files.

Saving text in registers

When you want to insert a copy of the same piece of text several times, it may be inconvenient to yank it from the kill ring, as each subsequent kill moves that entry further down the ring. An alternative is to store the text in a register and later retrieve it.

C-x r s r stores a copy of the text of the region into the register named r. If the mark is inactive, Emacs first reactivates the mark where it was last set. The mark is deactivated at the end of this command. See Mark. C-u C-x r s r, the same command with a prefix argument, copies the text into register r and deletes the text from the buffer as well; you can think of this as "moving" the region text into the register.

M-x append-to-register <RET> r appends the copy of the text in the region to the text already stored in the register named r. If invoked with a prefix argument, it deletes the region after appending it to the register. The command prepend-to-register is similar, except that it prepends the region text to the text in the register instead of appending it.

When you are collecting text using append-to-register and prepend-to-register, you may want to separate individual collected pieces using a separator. In that case, configure a register-separator and store the separator text in to that register. For example, to get double newlines as text separator during the collection process, you can use the following setting.

(setq register-separator ?+) (set-register register-separator "\n\n")

C-x r i r inserts in the buffer the text from register r. Normally it leaves point before the text and sets the mark after, without activating it. With a numeric argument, it instead puts point after the text and the mark before.

Saving rectangles in registers

A register can contain a rectangle instead of linear text. See Rectangles, for basic information on how to specify a rectangle in the buffer.

The C-x r i r (insert-register) command, previously documented in Text Registers, inserts a rectangle rather than a text string, if the register contains a rectangle.

Saving window configuration in registers

You can save the window configuration of the selected frame in a register, or even the configuration of all windows in all frames, and restore the configuration later. See Windows, for information about window configurations.

Use C-x r j r to restore a window or frame configuration. This is the same command used to restore a cursor position. When you restore a frame configuration, any existing frames not included in the configuration become invisible. To delete these frames instead, use C-u C-x r j r.

Keeping numbers in registers

There are commands to store a number in a register, to insert the number in the buffer in decimal, and to increment it. These commands can be useful in keyboard macros (see Keyboard Macros).

C-x r i is the same command used to insert any other sort of register contents into the buffer. C-x r + with no numeric argument increments the register value by 1; C-x r n with no numeric argument stores zero in the register.

Keeping file names in registers

If you visit certain file names frequently, you can visit them more conveniently if you put their names in registers. Here's the Lisp code used to put a file name in a register:

(set-register ?r '(file . name))

For example,

(set-register ?z '(file . "/gd/gnu/emacs/19.0/src/ChangeLog"))

Puts the file name shown in register 'z'.

To visit the file whose name is in register r, type C-x r j r. (This is the same command used to jump to a position or restore a frame configuration.)

Bookmarks

Bookmarks are somewhat like registers in that they record positions where you can jump. Unlike registers, they have long names, and they persist automatically from one Emacs session to the next. The prototypical use of bookmarks is to record "where you were reading" in various files.

The prototypical use for bookmarks is to record one current position in each of several files. So the command C-x r m, which sets a bookmark, uses the visited file name as the default for the bookmark name. If you name each bookmark after the file it points to, then you can conveniently revisit any of those files with C-x r b, and move to the position of the bookmark at the same time.

To display a list of all your bookmarks in a separate buffer, type C-x r l (list-bookmarks). If you switch to that buffer, you can use it to edit your bookmark definitions or annotate the bookmarks. Type C-h m in the bookmark buffer for more information about its special editing commands.

When you kill Emacs, Emacs saves your bookmarks, if you have changed any bookmark values. You can also save the bookmarks at any time with the M-x bookmark-save command. Bookmarks are saved to the file ~/.emacs.d/bookmarks (for compatibility with older versions of Emacs, if you have a file named ~/.emacs.bmk, that is used instead). The bookmark commands load your default bookmark file automatically. This saving and loading is how bookmarks persist from one Emacs session to the next.

If you set the variable bookmark-save-flag to 1, each command that sets a bookmark also saves your bookmarks; this way, you don't lose any bookmark values even if Emacs crashes. The value, if a number, says how many bookmark modifications should go by between saving. If you set this variable to nil, Emacs only saves bookmarks if you explicitly use M-x bookmark-save.

Bookmark position values are saved with surrounding context, so that bookmark-jump can find the proper position even if the file is modified slightly. The variable bookmark-search-size says how many characters of context to record on each side of the bookmark's position.

Here are some additional commands for working with bookmarks:

Controlling the display

Since only part of a large buffer fits in the window, Emacs has to show only a part of it. This chapter describes commands and variables that let you specify which part of the text you want to see, and how the text is displayed.

• Scrolling: Commands to move text up and down in a window.
• Recentering: A scroll command that centers the current line.
• Auto Scrolling: Redisplay scrolls text automatically when needed.
• Horizontal Scrolling: Moving text left and right in a window.
• Narrowing: Restricting display and editing to a portion of the buffer.
• View Mode: Viewing read-only buffers.
• Follow Mode: Follow mode lets two windows scroll as one.
• Faces: How to change the display style using faces.
• Colors: Specifying colors for faces.
• Standard Faces: The main predefined faces.
• Text Scale: Increasing or decreasing text size in a buffer.
• Font Lock: Minor mode for syntactic highlighting using faces.
• Highlight Interactively: Tell Emacs what text to highlight.
• Fringes: Enabling or disabling window fringes.
• Displaying Boundaries: Displaying top and bottom of the buffer.
• Useless Whitespace: Showing possibly spurious trailing whitespace.
• Selective Display: Hiding lines with lots of indentation.
• Optional Mode Line: Optional mode line display features.
• Text Display: How text characters are normally displayed.
• Cursor Display: Features for displaying the cursor.
• Line Truncation: Truncating lines to fit the screen width instead of continuing them to multiple screen lines.
• Visual Line Mode: Word wrap and screen line-based editing.
• Display Custom: Information on variables for customizing display.

Scrolling

If a window is too small to display all the text in its buffer, it displays only a portion of it. Scrolling commands change which portion of the buffer is displayed.

Scrolling "forward" or "up" advances the portion of the buffer displayed in the window; equivalently, it moves the buffer text upwards relative to the window. Scrolling "backward" or "down" displays an earlier portion of the buffer, and moves the text downwards relative to the window.

In Emacs, scrolling "up" or "down" refers to the direction that the text moves in the window, not the direction that the window moves relative to the text. This terminology was adopted by Emacs before the modern meaning of "scrolling up" and "scrolling down" became widespread. Hence, the strange result that <PageDown> scrolls "up" in the Emacs sense.

The portion of a buffer displayed in a window always contains point. If you move point past the bottom or top of the window, scrolling occurs automatically to bring it back onscreen (see Auto Scrolling). You can also scroll explicitly with these commands:

The number of lines of overlap left by these scroll commands is controlled by the variable next-screen-context-lines, whose default value is 2. You can supply the commands with a numeric prefix argument, n, to scroll by n lines; Emacs attempts to leave point unchanged, so that the text and point move up or down together. C-v with a negative argument is like M-v and vice versa.

By default, these commands signal an error (by beeping or flashing the screen) if no more scrolling is possible, because the window has reached the beginning or end of the buffer. If you change the variable scroll-error-top-bottom to t, the command moves point to the farthest possible position. If point is already there, the command signals an error.

Some users like scroll commands to keep point at the same screen position, so that scrolling back to the same screen conveniently returns point to its original position. You can enable this behavior via the variable scroll-preserve-screen-position. If the value is t, Emacs adonlys point to keep the cursor at the same screen position whenever a scroll command moves it off-window, rather than moving it to the topmost or bottommost line. With any other non-nil value, Emacs adonlys point this way even if the scroll command leaves point in the window. This variable affects all the scroll commands documented in this section, and scrolling with the mouse wheel (see Mouse Commands); in general, it affects any command with a non-nil scroll-command property.

The commands M-x scroll-up and M-x scroll-down behave similarly to scroll-up-command and scroll-down-command, except they do not obey scroll-error-top-bottom. Before Emacs 24, these were the default commands for scrolling up and down. The commands M-x scroll-up-line and M-x scroll-down-line scroll the current window by one line at a time. If you intend to use any of these commands, you might want to give them key bindings (see Init Rebinding).

Recentering

The C-l (recenter-top-bottom) command recenters the selected window, scrolling it so that the current screen line is exactly in the center of the window, or as close to the center as possible.

Typing C-l twice in a row (C-l C-l) scrolls the window so that point is on the topmost screen line. Typing a third C-l scrolls the window so that point is on the bottom-most screen line. Each successive C-l cycles through these three positions.

You can change the cycling order by customizing the list variable recenter-positions. Each list element should be the symbol top, middle, or bottom, or a number; an integer means to move the line to the specified screen line, while a floating-point number between 0.0 and 1.0 specifies a percentage of the screen space from the top of the window. The default, (middle top bottom), is the cycling order described above. Furthermore, if you change the variable scroll-margin to a non-zero value n, C-l always leaves at least n screen lines between point and the top or bottom of the window (see Auto Scrolling).

You can also give C-l a prefix argument. A plain prefix argument, C-u C-l, recenters point. A positive argument n puts point n lines down from the top of the window. An argument of zero puts point on the topmost line. A negative argument -n puts point n lines from the bottom of the window. When given an argument, C-l does not clear the screen or cycle through different screen positions.

If the variable recenter-redisplay has a non-nil value, each invocation of C-l also clears and redisplays the screen; the special value tty (the default) says to do this on text-terminal frames only. Redisplaying is useful in case the screen becomes garbled for any reason (see Screen Garbled).

The more primitive command M-x recenter behaves like recenter-top-bottom, but does not cycle among screen positions.

C-M-l (reposition-window) scrolls the current window heuristically in a way designed to get useful information onto the screen. For example, in a Lisp file, this command tries to get the entire current defun onto the screen if possible.

Automatic scrolling

Emacs performs automatic scrolling when point moves out of the visible portion of the text. Normally, automatic scrolling centers point vertically in the window, but there's several ways to alter this behavior.

If you set scroll-conservatively to a small number n, then moving point only a little off the screen (no more than n lines) causes Emacs to scroll only enough to bring point back on screen; if doing so fails to make point visible, Emacs scrolls only far enough to center point in the window. If you set scroll-conservatively to a large number (larger than 100), automatic scrolling never centers point, no matter how far point moves; Emacs always scrolls text only enough to bring point into view, either at the top or bottom of the window depending on the scroll direction. By default, scroll-conservatively is 0, which means to always center point in the window.

Another way to control automatic scrolling is to customize the variable scroll-step. Its value determines the number of lines by which to automatically scroll, when point moves off the screen. If scrolling by that number of lines fails to bring point back into view, point is centered instead. The default value is zero, which (by default) causes point to always be centered after scrolling.

A third way to control automatic scrolling is to customize the variables scroll-up-aggressively and scroll-down-aggressively, which directly specify the vertical position of point after scrolling. The value of scroll-up-aggressively should be either nil (the default), or a floating point number f between 0 and 1. The latter means that when point goes below the bottom window edge (i.e., scrolling forward), Emacs scrolls the window so that point is f parts of the window height from the bottom window edge. Thus, larger f means more aggressive scrolling: more new text is brought into view. The default value, nil, is equivalent to 0.5.

Likewise, scroll-down-aggressively is used when point goes above the bottom window edge (i.e., scrolling backward). The value specifies how far point should be from the top margin of the window after scrolling. Thus, as with scroll-up-aggressively, a larger value is more aggressive.

Note that the variables scroll-conservatively, scroll-step, and scroll-up-aggressively / scroll-down-aggressively control automatic scrolling in contradictory ways. Therefore, pick no more than one of these methods to customize automatic scrolling. In case you customize multiple variables, the order of priority is: scroll-conservatively, then scroll-step, and finally scroll-up-aggressively / scroll-down-aggressively.

The variable scroll-margin restricts how close point can come to the top or bottom of a window (even if aggressive scrolling specifies a fraction f that is larger than the window portion between the top and the bottom margins). Its value is some screen lines; if point comes within that many lines of the top or bottom of the window, Emacs performs automatic scrolling. By default, scroll-margin is 0.

Horizontal scrolling

Horizontal scrolling means shifting all the lines sideways within a window, so that some of the text near the left margin is not displayed. When the text in a window is scrolled horizontally, text lines are truncated rather than continued (see Line Truncation). If a window shows truncated lines, Emacs performs automatic horizontal scrolling whenever point moves off the left or right edge of the screen. To disable automatic horizontal scrolling, set the variable auto-hscroll-mode to nil. Note that when the automatic horizontal scrolling is turned off, if point moves off the edge of the screen, the cursor disappears to indicate that. (On text terminals, the cursor is left at the edge instead.)

The variable hscroll-margin controls how close point can get to the window's left and right edges before automatic scrolling occurs. It is measured in columns. For example, if the value is 5, then moving point within 5 columns of an edge causes horizontal scrolling away from that edge.

The variable hscroll-step determines how many columns to scroll the window when point gets too close to the edge. Zero, the default value, means to center point horizontally in the window. A positive integer value specifies the number of columns to scroll by. A floating-point number specifies the fraction of the window's width to scroll by.

You can also perform explicit horizontal scrolling with the following commands:

C-x < (scroll-left) scrolls text in the selected window to the left by the full width of the window, less two columns. (In other words, the text in the window moves left relative to the window.) With a numeric argument n, it scrolls by n columns.

If the text is scrolled to the left, and point moves off the left edge of the window, the cursor freezes at the left edge of the window, until point moves back to the displayed portion of the text. This is independent of the current setting of auto-hscroll-mode, which, for text scrolled to the left, only affects the behavior at the right edge of the window.

C-x > (scroll-right) scrolls similarly to the right. The window cannot be scrolled any farther to the right once it is displayed normally, with each line starting at the window's left margin; attempting to do so has no effect. This means that you don't have to calculate the argument precisely for C-x >; any sufficiently large argument restores the normal display.

If you use those commands to scroll a window horizontally, that sets a lower bound for automatic horizontal scrolling. Automatic scrolling continues to scroll the window, but never farther to the right than the amount you previously set by scroll-left.

Narrowing

Narrowing means focusing in on some portion of the buffer, making the rest temporarily inaccessible. The portion which you can still get to is called the accessible portion. Canceling the narrowing, which makes the entire buffer once again accessible, is called widening. The bounds of narrowing in effect in a buffer are called the buffer's restriction.

Narrowing can make it easier to concentrate on a single subroutine or paragraph by eliminating clutter. It can also be used to limit the range of operation of a replace command or repeating keyboard macro.

When you have narrowed down to a part of the buffer, that part appears to be all there is. You can't see the rest, you can't move into it (motion commands won't go outside the accessible part), you can't change it in any way. However, it is not gone, and if you save the file, all the inaccessible text is saved. The word 'Narrow' appears in the mode line whenever narrowing is in effect.

The primary narrowing command is C-x n n (narrow-to-region). It sets the current buffer's restrictions so that the text in the current region remains accessible, but all text before the region or after the region is inaccessible. Point and mark do not change.

Alternatively, use C-x n p (narrow-to-page) to reduce to the current page. See Pages, for the definition of a page. C-x n d (narrow-to-defun) narrows down to the defun containing point (see Defuns).

The way to cancel narrowing is to widen with C-x n w (widen). This makes all text in the buffer accessible again.

You can get information on what part of the buffer you are narrowed down to using the C-x = command. See Position Info.

Because narrowing can easily confuse users who do not understand it, narrow-to-region is normally a disabled command. Attempting to use this command asks for confirmation and gives you the option of enabling it; if you enable the command, confirmation is no longer required for it. See Disabling.

View mode

View mode is a minor mode that lets you scan a buffer by sequential screenfuls. It provides commands for scrolling through the buffer conveniently but not for changing it. Apart from the usual Emacs cursor motion commands, you can type <SPC> to scroll forward one windowful, <DEL> to scroll backward, and s to start an incremental search.

Typing q (View-quit) disables View mode, and switches back to the buffer and position before View mode was enabled. Typing e (View-exit) disables View mode, keeping the current buffer and position.

M-x view-buffer prompts for an existing Emacs buffer, switches to it, and enables View mode. M-x view-file prompts for a file and visits it with View mode enabled.

Follow mode

Follow mode is a minor mode that makes two windows, both showing the same buffer, scroll as a single tall "virtual window". To use Follow mode, go to a frame with only one window, split it into two side-by-side windows using C-x 3, and then type M-x follow-mode. From then on, you can edit the buffer in either of the two windows, or scroll either one; the other window follows it.

In Follow mode, if you move point outside the portion visible in one window and into the portion visible in the other window, that selects the other window; again, treating the two as if they were parts of one large window.

To turn off Follow mode, type M-x follow-mode a second time.

Text faces

Emacs can display text in different styles, called faces. Each face can specify various face attributes, such as the font, height, weight, slant, foreground and background color, and underlining or overlining. Most major modes assign faces to the text automatically, via Font Lock mode. See Font Lock, for more information about how these faces are assigned.

To see what faces are currently defined, and what they look like, type M-x list-faces-display. With a prefix argument, this prompts for a regular expression, and displays only faces with names matching that regular expression (see Regexps).

It's possible for a given face to look different in different frames. For instance, some text terminals do not support all face attributes, particularly font, height, and width, and some support a limited range of colors. Also, most Emacs faces are defined so that their attributes are different on light and dark frame backgrounds, for reasons of legibility. By default, Emacs automatically chooses which set of face attributes to display on each frame, based on the frame's current background color. However, you can override this by giving the variable frame-background-mode a non-nil value. A value of dark makes Emacs treat all frames as if they have a dark background, whereas a value of light makes it treat all frames as if they have a light background.

You can customize a face to alter its attributes, and save those customizations for future Emacs sessions. See Face Customization, for details.

The default face is the default for displaying text, and all its attributes are specified. Its background color is also used as the frame's background color. See Colors.

Another special face is the cursor face. On graphical displays, the background color of this face is used to draw the text cursor. None of the other attributes of this face have any effect; the foreground color for text under the cursor is taken from the background color of the underlying text. On text terminals, the appearance of the text cursor is determined by the terminal, not by the cursor face.

You can also use X resources to specify attributes of any particular face. See Resources.

Emacs can display variable-width fonts, but some Emacs commands, particularly indentation commands, do not account for variable character display widths. Therefore, we recommend not using variable-width fonts for most faces, particularly those assigned by Font Lock mode.

Colors for faces

Faces can have various foreground and background colors. When you specify a color for a face—for instance, when customizing the face (see Face Customization) you can use either a color name or an RGB triplet.

A color name is a pre-defined name, such as 'dark orange' or 'medium sea green'. To view a list of color names, type M-x list-colors-display. To control the order that colors are shown, customize list-colors-sort. If you run this command on a graphical display, it shows the full range of color names known to Emacs (these are the standard X11 color names, defined in X's rgb.txt file). If you run the command on a text terminal, it shows only a small subset of colors that are safely displayed on such terminals. However, Emacs understands X11 color names even on text terminals; if a face is given a color specified by an X11 color name, it is displayed using the closest-matching terminal color.

An RGB triplet is a string of the form '#RRGGBB'. Each of the R, G, and B components is a hexadecimal number specifying the component's relative intensity, one to four digits long (usually two digits are used). The components must have the same number of digits. For hexadecimal values A to F, either upper or lowercase are acceptable.

The M-x list-colors-display command also shows the equivalent RGB triplet for each named color. For instance, 'medium sea green' is equivalent to '#3CB371'.

You can change the foreground and background colors of a face with M-x set-face-foreground and M-x set-face-background. These commands prompt in the minibuffer for a face name and a color, with completion, and then set that face to use the specified color. They affect the face colors on all frames, but their effects do not persist for future Emacs sessions, unlike using the customization buffer or X resources. You can also use frame parameters to set foreground and background colors for a specific frame; See Frame Parameters.

Standard faces

Here are the standard faces for specifying text appearance. You can apply them to specific text when you want the effects they produce.

Here's an incomplete list of faces used to highlight parts of the text temporarily for specific purposes. (Other modes define their faces for this purpose.)

The following faces control the appearance of parts of the Emacs frame:

The following faces likewise control the appearance of parts of the Emacs frame, but only on text terminals, or when Emacs is built on X with no toolkit support. (For all other cases, the appearance of the respective frame elements is determined by system-wide settings.)

Text scale

To increase the height of the default face in the current buffer, type C-x C-+ or C-x C-=. To decrease it, type C-x C--. To restore the default (global) face height, type C-x C-0. These keys are all bound to the same command, text-scale-adonly, which looks at the last key typed to determine which action to take.

The final key of these commands may be repeated without the leading C-x. For instance, C-x C-= C-= C-= increases the face height by three steps. Each step scales the text height by a factor of 1.2; to change this factor, customize the variable text-scale-mode-step. As an exception, a numeric argument of 0 to the text-scale-adonly command restores the default height, similar to typing C-x C-0.

The commands text-scale-increase and text-scale-decrease increase or decrease the height of the default face, only like C-x C-+ and C-x C-- respectively. You may find it convenient to bind to these commands, rather than text-scale-adonly.

The command text-scale-set scales the height of the default face in the current buffer to an absolute level specified by its prefix argument. The above commands automatically enable the minor mode text-scale-mode if the current font scaling is other than 1, and disable it otherwise.

Font lock mode

Font Lock mode is a minor mode, always local to a particular buffer, which assigns faces to (or "fontifies") the text in the buffer. Each buffer's major mode tells Font Lock mode which text to fontify; for instance, programming language modes fontify syntactically relevant constructs like comments, strings, and function names.

Font Lock mode is enabled by default. To toggle it in the current buffer, type M-x font-lock-mode. A positive numeric argument unconditionally enables Font Lock mode, and a negative or zero argument disables it.

Type M-x global-font-lock-mode to toggle Font Lock mode in all buffers. To impose this setting for future Emacs sessions, customize the variable global-font-lock-mode (see Easy Customization), or add the following line to your init file:

(global-font-lock-mode 0)

If you have disabled Global Font Lock mode, you can still enable Font Lock for specific major modes by adding the function font-lock-mode to the mode hooks (see Hooks). For example, to enable Font Lock mode for editing C files, you can do this:

(add-hook 'c-mode-hook 'font-lock-mode)

Font Lock mode uses several specifically named faces to do its job, including font-lock-string-face, font-lock-comment-face, and others. The easiest way to find them all is to use M-x customize-group <RET> font-lock-faces <RET>. You can then use that customization buffer to customize the appearance of these faces. See Face Customization.

You can customize the variable font-lock-maximum-decoration to alter the amount of fontification applied by Font Lock mode, for major modes that support this feature. The value should be a number (with 1 representing a minimal amount of fontification; some modes support levels as high as 3); or t, meaning "as high as possible" (the default). You can also specify different numbers for particular major modes; for example, to use level 1 for C/C++ modes, and the default level otherwise, use the value

'((c-mode . 1) (c++-mode . 1)))

Comment and string fontification (or "syntactic" fontification) relies on analysis of the syntactic structure of the buffer text. For the sake of speed, some modes, including Lisp mode, rely on a special convention: an open-parenthesis or open-brace in the leftmost column always defines the beginning of a defun, and is thus always outside any string or comment. Therefore, avoid placing an open-parenthesis or open-brace in the leftmost column, if it's inside a string or comment. See Left Margin Paren, for details.

The variable font-lock-beginning-of-syntax-function, which is always buffer-local, specifies how Font Lock mode can find a position guaranteed to be outside any comment or string. In modes which use the leftmost column parenthesis convention, the default value of the variable is beginning-of-defun that tells Font Lock mode to use the convention. If you set this variable to nil, Font Lock no longer relies on the convention. This avoids incorrect results, but the price is that, in some cases, fontification for a changed text must rescan buffer text from the beginning of the buffer. This can considerably slow down redisplay while scrolling, particularly if you are close to the end of a large buffer.

Font Lock highlighting patterns already exist for most modes, but you may want to fontify additional patterns. You can use the function font-lock-add-keywords, to add highlighting patterns for a particular mode. For example, to highlight 'FIXME:' words in C comments, use this:

(add-hook 'c-mode-hook
	(lambda ()
		(font-lock-add-keywords
		 nil
		 '(("\\<\\(FIXME\\):"
		 1
		 font-lock-warning-face
		 t))
		)
	)
)

To remove keywords from the font-lock highlighting patterns, use the function font-lock-remove-keywords.

Fontifying large buffers can take a long time. To avoid large delays when a file is visited, Emacs initially fontifies only the visible portion of a buffer. As you scroll through the buffer, each portion that becomes visible is fontified as soon as it is displayed; this type of Font Lock is called Just-In-Time (or JIT) Lock. You can control how JIT Lock behaves, including telling it to perform fontification while idle, by customizing variables in the customization group 'jit-lock'. See Specific Customization.

Interactive highlighting

Highlight Changes mode is a minor mode that highlights the parts of the buffer that were changed most recently, by giving that text a different face. To enable or disable Highlight Changes mode, use M-x highlight-changes-mode.

"Hi Lock" mode is a minor mode that highlights text that matches regular expressions you specify. For example, you can use it to highlight all the references to a certain variable in a program source file, highlight certain parts in a voluminous output of some program, or highlight certain names in an article. To enable or disable Hi Lock mode, use the command M-x hi-lock-mode. To enable Hi Lock mode for all buffers, use M-x global-hi-lock-mode or place (global-hi-lock-mode 1) in your .emacs file.

Hi Lock mode works like Font Lock mode (see Font Lock), except that you specify explicitly the regular expressions to highlight. You control them with these commands:

Also, hi-lock-find-patterns does nothing if the current major mode's symbol is a member of the list hi-lock-exclude-modes.

Window fringes

On graphical displays, each Emacs window normally has narrow fringes on the left and right edges. The fringes are used to display symbols that provide information about the text in the window. You can type M-x fringe-mode to disable the fringes, or modify their width. This command affects fringes in all frames; to modify fringes on the selected frame only, use M-x set-fringe-style.

The most common use of the fringes is to indicate a continuation line (see Continuation Lines). When one line of text is split into multiple screen lines, the left fringe shows a curving arrow for each screen line except the first, indicating that "this is not the real beginning". The right fringe shows a curving arrow for each screen line except the last, indicating that "this is not the real end". If the line's direction is right-to-left (see Bidirectional Editing), the meanings of the curving arrows in the fringes are swapped.

The fringes indicate line truncation with short horizontal arrows meaning "there's more text on this line which is scrolled horizontally out of view". Clicking the mouse on one of the arrows scrolls the display horizontally in the direction of the arrow.

The fringes can also indicate other things, such as buffer boundaries (see Displaying Boundaries), and where a program you are debugging is executing (see Debuggers).

The fringe is also used for drawing the cursor, if the current line is exactly as wide as the window and point is at the end of the line. To disable this, change the variable overflow-newline-into-fringe to nil; this causes Emacs to continue or truncate lines that are exactly as wide as the window.

Displaying boundaries

On graphical displays, Emacs can indicate the buffer boundaries in the fringes. If you enable this feature, the first line and the last line are marked with angle images in the fringes. This can be combined with up and down arrow images which say whether it is possible to scroll the window.

The buffer-local variable indicate-buffer-boundaries controls how the buffer boundaries and window scrolling is indicated in the fringes. If the value is left or right, both angle and arrow bitmaps are displayed in the left or right fringe, respectively.

If value is an alist, each element (indicator . position) specifies the position of one of the indicators. The indicator must be one of top, bottom, up, down, or t which specifies the default position for the indicators not present in the alist. The position is one of left, right, or nil which specifies not to show this indicator.

For example, ((top . left) (t . right)) places the top angle bitmap in left fringe, the bottom angle bitmap in right fringe, and both arrow bitmaps in right fringe. To show only the angle bitmaps in the left fringe, but no arrow bitmaps, use ((top . left) (bottom . left)).

Useless whitespace

It is easy to leave unnecessary spaces at the end of a line, or empty lines at the end of a buffer, without realizing it. In most cases, this trailing whitespace has no effect, but sometimes it is a nuisance.

You can make trailing whitespace at the end of a line visible by setting the buffer-local variable show-trailing-whitespace to t. Then Emacs displays trailing whitespace, using the face trailing-whitespace.

This feature does not apply when point is at the end of the line containing the whitespace. Strictly speaking, that is "trailing whitespace" nonetheless, but displaying it specially in that case looks ugly while you are typing in new text. In this special case, the location of point is enough to show you that the spaces are present.

Type M-x delete-trailing-whitespace to delete all trailing whitespace. This command deletes all extra spaces at the end of each line in the buffer, and all empty lines at the end of the buffer; to ignore the latter, change the variable delete-trailing-lines to nil. If the region is active, the command instead deletes extra spaces at the end of each line in the region.

On graphical displays, Emacs can indicate unused lines at the end of the window with a small image in the left fringe (see Fringes). The image appears for screen lines that do not correspond to any buffer text, so blank lines at the end of the buffer stand out because they lack this image. To enable this feature, set the buffer-local variable indicate-empty-lines to a non-nil value. You can enable or disable this feature for all new buffers by setting the default value of this variable, e.g., (setq-default indicate-empty-lines t).

Whitespace mode is a buffer-local minor mode that lets you "visualize" many kinds of whitespace in the buffer, by either drawing the whitespace characters with a special face or displaying them as special glyphs. To toggle this mode, type M-x whitespace-mode. The kinds of whitespace visualized are determined by the list variable whitespace-style. Here is a partial list of possible elements (see the variable's documentation for the full list):

Selective display

Emacs can hide lines indented more than a given number of columns. You can use this to get an overview of a part of a program.

To hide lines in the current buffer, type C-x $ (set-selective-display) with a numeric argument n. Then lines with at least n columns of indentation disappear from the screen. The only indication of their presence is that three dots ('...') appear at the end of each visible line that is followed by one or more hidden ones.

The commands C-n and C-p move across the hidden lines as if they were not there.

The hidden lines are still present in the buffer, and most editing commands see them as usual, so you may find point in the middle of the hidden text. When this happens, the cursor appears at the end of the previous line, after the three dots. If point is at the end of the visible line, before the newline that ends it, the cursor appears before the three dots.

To make all lines visible again, type C-x $ with no argument.

If you set the variable selective-display-ellipses to nil, the three dots do not appear at the end of a line that precedes hidden lines. Then there is no visible indication of the hidden lines. This variable becomes local automatically when set.

See also Outline Mode for another way to hide part of the text in a buffer.

Optional mode line features

The buffer percentage pos indicates the percentage of the buffer above the top of the window. You can additionally display the size of the buffer by typing M-x size-indication-mode to turn on Size Indication mode. The size is displayed immediately following the buffer percentage like this:

POS of SIZE

Here SIZE is the human readable representation of the number of characters in the buffer, which means that k for 10^3, M for 10^6, G for 10^9, etc., are used to abbreviate.

The current line number of point appears in the mode line when Line Number mode is enabled. Use the command M-x line-number-mode to turn this mode on and off; normally it is on, by default. The line number appears after the buffer percentage pos, with the letter 'L' to indicate what it is.

Similarly, you can display the current column number by turning on Column number mode with M-x column-number-mode. The column number is indicated by the letter 'C'. However, when both of these modes are enabled, the line and column numbers are displayed in parentheses, the line number first, rather than with 'L' and 'C'. For example: '(561,2)'. See Minor Modes, for more information about minor modes and about how to use these commands.

If you have narrowed the buffer (see Narrowing), the displayed line number is relative to the accessible portion of the buffer. Thus, it isn't suitable as an argument to goto-line. (Use what-line command to see the line number relative to the whole file.)

If the buffer is very large (larger than the value of line-number-display-limit), Emacs won't compute the line number, because that would be too slow; therefore, the line number won't appear on the mode-line. To remove this limit, set line-number-display-limit to nil.

Line-number computation can also be slow if the lines in the buffer are too long. For this reason, Emacs doesn't display line numbers if the average width, in characters, of lines near point is larger than the value of line-number-display-limit-width. The default value is 200 characters.

Emacs can optionally display the time and system load in all mode lines. To enable this feature, type M-x display-time or customize the option display-time-mode. The information added to the mode line looks like this:

hh:mmpm l.ll

Here hh and mm are the hour and minute, followed always by 'am' or 'pm'. l.ll is the average number, collected for the last few minutes, of processes in the whole system that were either running or ready to run (i.e., were waiting for an available processor). (Some fields may be missing if your operating system cannot support them.) If you prefer time display in 24-hour format, set the variable display-time-24hr-format to t.

The word 'Mail' appears after the load level if there is mail for you that you have not read yet. On graphical displays, you can use an icon instead of 'Mail' by customizing display-time-use-mail-icon; this may save some space on the mode line. You can customize display-time-mail-face to make the mail indicator prominent. Use display-time-mail-file to specify the mail file to check, or set display-time-mail-directory to specify the directory to check for incoming mail (any nonempty regular file in the directory is considered as "newly arrived mail").

When running Emacs on a laptop computer, you can display the battery charge on the mode-line, using the command display-battery-mode or customizing the variable display-battery-mode. The variable battery-mode-line-format determines the way the battery charge is displayed; the exact mode-line message depends on the operating system, and it usually shows the current battery charge as a percentage of the total charge.

On graphical displays, the mode line is drawn as a 3D box. If you don't like this effect, you can disable it by customizing the mode-line face and setting its box attribute to nil. See Face Customization.

By default, the mode line of non-selected windows is displayed in a different face, called mode-line-inactive. Only the selected window is displayed in the mode-line face. This helps show which window is selected. When the minibuffer is selected, because it has no mode line, the window where you activated the minibuffer has its mode line displayed using mode-line; as a result, ordinary entry to the minibuffer does not change any mode lines.

You can disable use of mode-line-inactive by setting variable mode-line-in-non-selected-windows to nil; then all mode lines are displayed in the mode-line face.

You can customize the mode line display for each of the end-of-line formats by setting each of the variables eol-mnemonic-unix, eol-mnemonic-dos, eol-mnemonic-mac, and eol-mnemonic-undecided to the strings you prefer.

How text is displayed

Most characters are printing characters: when they appear in a buffer, they are displayed literally on the screen. Printing characters include ASCII numbers, letters, and punctuation characters, and many non-ASCII characters.

The ASCII character set contains non-printing control characters. Two of these are displayed specially: the newline character (Unicode code point U+000A) is displayed by starting a new line, while the tab character (U+0009) is displayed as a space that extends to the next tab stop column (normally every 8 columns). The number of spaces per tab is controlled by the buffer-local variable tab-width, which must have an integer value between 1 and 1000, inclusive. Note that how the tab character in the buffer is displayed has nothing to do with the definition of <TAB> as a command.

Other ASCII control characters, whose codes are below U+0020 (octal 40, decimal 32), are displayed as a caret ('^') followed by the non-control version of the character, with the escape-glyph face. For instance, the 'control-A' character, U+0001, is displayed as '^A'.

The raw bytes with codes U+0080 (octal 200) through U+009F (octal 237) are displayed as octal escape sequences, with the escape-glyph face. For instance, character code U+0098 (octal 230) is displayed as '\230'. If you change the buffer-local variable ctl-arrow to nil, the ASCII control characters are also displayed as octal escape sequences instead of caret escape sequences.

Some non-ASCII characters have the same appearance as an ASCII space or hyphen (minus) character. Such characters can cause problems if they are entered into a buffer without your realization, e.g., by yanking; for instance, source code compilers often do not treat non-ASCII spaces as whitespace characters. To deal with this problem, Emacs displays such characters specially: it displays U+00A0 (no-break space) with the nobreak-space face, and it displays U+00AD (soft hyphen), U+2010 (hyphen), and U+2011 (non-breaking hyphen) with the escape-glyph face. To disable this, change the variable nobreak-char-display to nil. If you give this variable a non-nil and non-t value, Emacs instead displays such characters as a highlighted backslash followed by a space or hyphen.

You can customize the way any particular character code is displayed by means of a display table.

On graphical displays, some characters may have no glyphs in any of the fonts available to Emacs. These glyphless characters are normally displayed as boxes containing the hexadecimal character code. Similarly, on text terminals, characters that are not displayed using the terminal encoding (see Terminal Coding) are normally displayed as question signs. You can control the display method by customizing the variable glyphless-char-display-control.

Displaying the cursor

On a text terminal, the cursor's appearance is controlled by the terminal, largely out of the control of Emacs. Some terminals offer two different cursors: a "visible" static cursor, and a "very visible" blinking cursor. By default, Emacs uses the very visible cursor, and switches to it when you start or resume Emacs. If the variable visible-cursor is nil when Emacs starts or resumes, it uses the normal cursor.

On a graphical display, many more properties of the text cursor is altered. To customize its color, change the :background attribute of the face named cursor (see Face Customization). (The other attributes of this face have no effect; the text shown under the cursor is drawn using the frame's background color.) To change its shape, customize the buffer-local variable cursor-type; possible values are box (the default), hollow (a hollow box), bar (a vertical bar), (bar . n) (a vertical bar n pixels wide), hbar (a horizontal bar), (hbar . n) (a horizontal bar n pixels tall), or nil (no cursor at all).

To disable cursor blinking, change the variable blink-cursor-mode to nil (see Easy Customization), or add the line (blink-cursor-mode 0) to your init file. Alternatively, you can change how the cursor looks when it "blinks off" by customizing the list variable blink-cursor-alist. Each element in the list should have the form (on-type . off-type); this means that if the cursor is displayed as on-type when it blinks on (where on-type is one of the cursor types described above), then it is displayed as off-type when it blinks off.

Some characters, such as tab characters, are "extra wide". When the cursor is positioned over such a character, it is normally drawn with the default character width. You can make the cursor stretch to cover wide characters, by changing the variable x-stretch-cursor to a non-nil value.

The cursor normally appears in non-selected windows as a non-blinking hollow box. (For a bar cursor, it instead appears as a thinner bar.) To turn off cursors in non-selected windows, change the variable cursor-in-non-selected-windows to nil.

To make the cursor even more visible, you can use HL Line mode, a minor mode that highlights the line containing point. Use M-x hl-line-mode to enable or disable it in the current buffer. M-x global-hl-line-mode enables or disables the same mode globally.

Line truncation

As an alternative to continuation (see Continuation Lines), Emacs can display long lines by truncation. This means that all the characters that do not fit in the width of the screen or window do not appear at all. On graphical displays, a small straight arrow in the fringe indicates truncation at either end of the line. On text terminals, this is indicated with '$' signs in the leftmost and/or rightmost columns.

Horizontal scrolling automatically causes line truncation (see Horizontal Scrolling). You can explicitly enable line truncation for a particular buffer with the command M-x toggle-truncate-lines. This works by locally changing the variable truncate-lines. If that variable is non-nil, long lines are truncated; if it's nil, they are continued onto multiple screen lines. Setting the variable truncate-lines in any way makes it local to the current buffer; until that time, the default value, which is normally nil, is in effect.

If a split window becomes too narrow, Emacs may automatically enable line truncation. See Split Window, for the variable truncate-partial-width-windows which controls this.

Visual line mode

Another alternative to ordinary line continuation is to use word wrap. Here, each long logical line is divided into two or more screen lines, like in ordinary line continuation. However, Emacs attempts to wrap the line at word boundaries near the right window edge. This makes the text easier to read, as wrapping does not occur in the middle of words.

Word wrap is enabled by Visual Line mode, an optional minor mode. To turn on Visual Line mode in the current buffer, type M-x visual-line-mode; repeating this command turns it off. You can also turn on Visual Line mode using the menu bar: in the Options menu, select the 'Line Wrapping in this Buffer' submenu, followed by the 'Word Wrap (Visual Line Mode)' menu item. While Visual Line mode is enabled, the mode-line shows the string 'wrap' in the mode display. The command M-x global-visual-line-mode toggles Visual Line mode in all buffers.

In Visual Line mode, some editing commands work on-screen lines instead of logical lines: C-a (beginning-of-visual-line) moves to the beginning of the screen line, C-e (end-of-visual-line) moves to the end of the screen line, and C-k (kill-visual-line) kills text to the end of the screen line.

To move by logical lines, use the commands M-x next-logical-line and M-x previous-logical-line. These move point to the next logical line and the previous logical line respectively, regardless of whether Visual Line mode is enabled. If you use these commands frequently, it may be convenient to assign key bindings to them. See Init Rebinding.

By default, word-wrapped lines do not display fringe indicators. Visual Line mode is often used to edit files that contain many long logical lines, so having a fringe indicator for each wrapped line would be visually distracting. You can change this by customizing the variable visual-line-fringe-indicators.

Customization of display

This section describes variables that control miscellaneous aspects of the appearance of the Emacs screen. Beginning users can skip it.

If the variable visible-bell is non-nil, Emacs attempts to make the whole screen blink when it would normally make an audible bell sound. This variable has no effect if your terminal does not have a way to make the screen blink.

The variable echo-keystrokes controls the echoing of multi-character keys; its value is the number of seconds of pause required to cause echoing to start, or zero, meaning don't echo at all. The value takes effect when there is something to echo. See Echo Area.

On graphical displays, Emacs displays the mouse pointer as an hourglass if Emacs is busy. To disable this feature, set the variable display-hourglass to nil. The variable hourglass-delay determines the number of seconds of "busy time" before the hourglass is shown; the default is 1.

If the mouse pointer lies inside an Emacs frame, Emacs makes it invisible each time you type a character to insert text, to prevent it from obscuring the text. (To be precise, the hiding occurs when you type a "self-inserting" character. See Inserting Text.) Moving the mouse pointer makes it visible again. To disable this feature, set the variable make-pointer-invisible to nil.

On graphical displays, the variable underline-minimum-offset determines the minimum distance between the baseline and underline, in pixels, for underlined text. By default, the value is 1; increasing it may improve the legibility of underlined text for certain fonts. (However, Emacs never draws the underline below the current line area.) The variable x-underline-at-descent-line determines how to draw underlined text. The default is nil, which means to draw it at the baseline level of the font; if you change it to nil, Emacs draws the underline at the same height as the font's descent line.

The variable overline-margin specifies the vertical position of an overline above the text, including the height of the overline itself, in pixels; the default is 2.

On some text terminals, bold face and inverse video together result in text that is hard to read. Call the function tty-suppress-bold-inverse-default-colors with a non-nil argument to suppress the effect of bold-face in this case.

Searching and replacement

Like other editors, Emacs has commands to search for occurrences of a string. Emacs also has commands to replace occurrences of a string with a different string. There are also commands that do the same thing, but search for patterns instead of fixed strings.

You can also search multiple files under the control of a tags table (see Tags Search) or through the Dired A command (see Operating on Files), or ask the grep program to do it (see Grep Searching).

• Incremental Search: Search happens as you type the string.
• Nonincremental Search: Specify entire string and then search.
• Word Search: Search for sequence of words.
• Symbol Search: Search for a source code symbol.
• Regexp Search: Search for match for a regexp.
• Regexps: Syntax of regular expressions.
• Regexp Backslash: Regular expression constructs starting with `\'.
• Regexp Example: A complex regular expression explained.
• Search Case: To ignore case while searching, or not.
• Replace: Search, and replace some or all matches.
• Other Repeating Search: Operating on all matches for some regexp.

Incremental search

The principal search command in Emacs is incremental: it begins searching as soon as you type the first character of the search string. As you type in the search string, Emacs shows you where the string (as you have typed it so far) would be found. When you have typed enough characters to identify the place you want, you can stop. Depending on what you plan to do next, you may or may not need to terminate the search explicitly with <RET>.

• Basic Isearch: Basic incremental search commands.
• Repeat Isearch: Searching for the same string again.
• Error in Isearch: When your string is not found.
• Special Isearch: Special input in incremental search.
• Isearch Yank: Commands that grab text into the search string or else edit the search string.
• Isearch Scroll: Scrolling during an incremental search.
• Isearch Minibuffer: Incremental search of the minibuffer history.

Basics of incremental search

C-s (isearch-forward) starts a forward incremental search. It reads characters from the keyboard, and moves point only past the end of the next occurrence of those characters in the buffer.

For instance, if you type C-s and then F, that puts the cursor after the first 'F' that occurs in the buffer after the starting point. Then if you then type O, the cursor moves to only after the first 'FO'; the 'F' in that 'FO' might not be the first 'F' previously found. After another O, the cursor moves to only after the first 'FOO'.

At each step, Emacs highlights the current match -- the buffer text that matches the search string -- using the isearch face (see Faces). The current search string is also displayed in the echo area.

If you make a mistake typing the search string, type <DEL>. Each <DEL> cancels the last character of the search string.

When you are satisfied with the place you have reached, type <RET>. This stops searching, leaving the cursor where the search brought it. Also, any command not specially meaningful in searches stops the searching and is then executed. Thus, typing C-a exits the search and then moves to the beginning of the line. <RET> is necessary only if the next command you want to type is a printing character, <DEL>, <RET>, or another character that is special within searches (C-q, C-w, C-r, C-s, C-y, M-y, M-r, M-c, M-e, and some others described below).

As a special exception, entering <RET> when the search string is empty launches nonincremental search (see Nonincremental Search).

When you exit the incremental search, it adds the original value of point to the mark ring, without activating the mark; you can thus use C-u C-<SPC> to return to where you were before beginning the search. See Mark Ring. It only does this if the mark was not already active.

To search backwards, use C-r (isearch-backward) instead of C-s to start the search. A backward search finds matches that end before the starting point, only as a forward search finds matches that begin after it.

Repeating incremental search

Suppose you search forward for 'FOO' and find a match, but not the one you expected to find: the 'FOO' you were aiming for occurs later in the buffer. In this event, type another C-s to move to the next occurrence of the search string. You can repeat this any number of times. If you overshoot, you can cancel some C-s characters with <DEL>. Similarly, each C-r in a backward incremental search repeats the backward search.

If you pause for a little while during incremental search, Emacs highlights all the other possible matches for the search string that are present on the screen. This helps you anticipate where you can get to by typing C-s or C-r to repeat the search. The other matches are highlighted differently from the current match, using the customizable face lazy-highlight (see Faces). If you don't like this feature, you can disable it by setting isearch-lazy-highlight to nil.

After exiting a search, you can search for the same string again by typing only C-s C-s. The first C-s is the key that invokes incremental search, and the second C-s means "search again". Similarly, C-r C-r searches backward for the last search string. In determining the last search string, it doesn't matter whether the string was searched for with C-s or C-r.

If you are searching forward but you realize you were looking for something before the starting point, type C-r to switch to a backward search, leaving the search string unchanged. Similarly, C-s in a backward search switches to a forward search.

If a search is failing and you ask to repeat it by typing another C-s, it starts again from the beginning of the buffer. Repeating a failing reverse search with C-r starts again from the end. This is called wrapping around, and 'Wrapped' appears in the search prompt once this has happened. If you keep on going past the original starting point of the search, it changes to 'Overwrapped', which means that you are revisiting matches that you have already seen.

To reuse earlier search strings, use the search ring. The commands M-p and M-n move through the ring to pick a search string to reuse. These commands leave the selected search ring element in the minibuffer, where you can edit it.

To edit the current search string in the minibuffer without replacing it with items from the search ring, type M-e. Type C-s or C-r to finish editing the string and search for it.

Errors in incremental Isearch

If your string is not found at all, the echo area says 'Failing I-Search', and the cursor moves past the place where Emacs found as much of your string as it could. Thus, if you search for 'FOOT', and there is no 'FOOT', you might see the cursor after the 'FOO' in 'FOOL'. In the echo area, the part of the search string that failed to match is highlighted using the face isearch-fail.

At this point, there's several things you can do. If your string was mistyped, you can use <DEL> to erase some of it and correct it. If you like the place you have found, you can type <RET> to remain there. Or you can type C-g, which removes from the search string the characters that could not be found (the 'T' in 'FOOT'), leaving those that were found (the 'FOO' in 'FOOT'). A second C-g at that point cancels the search entirely, returning point to where it was when the search started.

The quit command, C-g, does special things during searches; only what it does depends on the status of the search. If the search has found what you specified and is waiting for input, C-g cancels the entire search, moving the cursor back to where you started the search. If C-g is typed when there are characters in the search string that have not been found (because Emacs is still searching for them, or because it has failed to find them), then the search string characters which have not been found are discarded from the search string. With them gone, the search is now successful and waiting for more input, so a second C-g will cancel the entire search.

Special input for incremental search

Some of the characters you type during incremental search have special effects.

By default, incremental search performs "lax" space matching: each space, or sequence of spaces, matches any sequence of one or more spaces in the text. Hence, 'foo bar' matches 'foo bar', 'foo bar', 'foo bar', and so on (but not 'foobar'). More precisely, Emacs matches each sequence of space characters in the search string to a regular expression specified by the variable search-whitespace-regexp. For example, set it to '"[[:space:]\n]+"' to make spaces match sequences of newlines and spaces. To toggle lax space matching, type M-s <SPC> (isearch-toggle-lax-whitespace). To disable this feature entirely, change search-whitespace-regexp to nil; then each space in the search string matches exactly one space.

If the search string you entered contains only lower-case letters, the search is case-insensitive; as long as an upper-case letter exists in the search string, the search becomes case-sensitive. If you delete the upper-case character from the search string, it ceases to have this effect. See Search Case.

To search for a newline character, type C-j.

To search for other control characters, such as <control-S>, quote it by typing C-q first (see Inserting Text). To search for non-ASCII characters, you can either use C-q and enter its octal code, or use an input method (see Input Methods). If an input method is enabled in the current buffer when you start the search, you can use it in the search string also. While typing the search string, you can toggle the input method with the command C-\ (isearch-toggle-input-method). You can also turn on a non-default input method with C-^ (isearch-toggle-specified-input-method), which prompts for the name of the input method. When an input method is active during incremental search, the search prompt includes the input method mnemonic, like this:

I-search [im]:

Where im is the mnemonic of the active input method. Any input method you enable during incremental search remains enabled in the current buffer afterwards.

Typing M-% in incremental search invokes query-replace or query-replace-regexp (depending on search mode) with the current search string used as the string to replace. See Query Replace.

Typing M-<TAB> in incremental search invokes isearch-complete, which attempts to complete the search string using the search ring as a list of completion alternatives. See Completion. In many operating systems, the M-<TAB> key sequence is captured by the window manager; you then need to rebind isearch-complete to another key sequence if you want to use it (see Rebinding).

When incremental search is active, you can type C-h C-h to access interactive help options, including a list of special key bindings. These key bindings are part of the keymap isearch-mode-map (see Keymaps).

Isearch yanking

In incremental search, C-y (isearch-yank-kill) appends the current kill to the search string. M-y (isearch-yank-pop), if called after C-y, replaces that appended text with an earlier kill, similar to the usual M-y (yank-pop) command (see Yanking). Mouse-2 appends the current X selection (see Primary Selection).

C-w (isearch-yank-word-or-char) appends the next character or word at point to the search string. This is an easy way to search for another occurrence of the text at point. (The decision of whether to copy a character or a word is heuristic.)

Similarly, M-s C-e (isearch-yank-line) appends the rest of the current line to the search string. If point is already at the end of a line, it appends the next line.

If the search is currently case-insensitive, both C-w and M-s C-e convert the text they copy to lowercase, so that the search remains case-insensitive.

C-M-w (isearch-del-char) deletes the last character from the search string, and C-M-y (isearch-yank-char) appends the character after point to the search string. An alternative method to add the character after point is to enter the minibuffer with M-e (see Repeat Isearch) and type C-f at the end of the search string in the minibuffer.

Scrolling during incremental search

Normally, scrolling commands exit incremental search. If you change the variable isearch-allow-scroll to a non-nil value, that enables the use of the scroll-bar, and keyboard scrolling commands like C-v, M-v, and C-l (see Scrolling). This applies only to calling these commands via their bound key sequences; typing M-x still exits the search. You can give prefix arguments to these commands in the usual way. This feature won't let you scroll the current match out of visibility, however.

The isearch-allow-scroll feature also affects some other commands, such as C-x 2 (split-window-below) and C-x ^ (enlarge-window), which don't exactly scroll but do affect where the text appears on the screen. It applies to any command whose name has a non-nil isearch-scroll property. So you can control which commands are affected by changing these properties.

For example, to make C-h l usable within an incremental search in all future Emacs sessions, use C-h c to find what command it runs (see Key Help), which is view-lossage. Then you can put the following line in your init file (see Init File):

(put 'view-lossage 'isearch-scroll t)

This feature is applied to any command that doesn't permanently change point, the buffer contents, the match data, the current buffer, or the selected window and frame. The command must not itself attempt an incremental search.

Searching the minibuffer

If you start an incremental search while the minibuffer is active, Emacs searches the contents of the minibuffer. Unlike searching an ordinary buffer, the search string is not shown in the echo area, because that is used to display the minibuffer.

If an incremental search fails in the minibuffer, it tries searching the minibuffer history. See Minibuffer History. You can visualize the minibuffer and its history as a series of "pages", with the earliest history element on the first page and the current minibuffer on the last page. A forward search, C-s, searches forward to later pages; a reverse search, C-r, searches backwards to earlier pages. Like in ordinary buffer search, a failing search can wrap around, going from the last page to the first page or vice versa.

When the current match is on a history element, that history element is pulled into the minibuffer. If you exit the incremental search normally (e.g., by typing <RET>), it remains in the minibuffer afterwards. Canceling the search, with C-g, restores the contents of the minibuffer when you began the search.

Non-incremental search

Emacs also has conventional nonincremental search commands, which require you to type the entire search string before searching begins.

To start a nonincremental search, first type C-s <RET>. This enters the minibuffer to read the search string; terminate the string with <RET>, and then the search takes place. If the string is not found, the search command signals an error.

When you type C-s <RET>, the C-s invokes incremental search as usual. That command is specially programmed to invoke the command for nonincremental search, search-forward, if the string you specify is empty. (Such an empty argument would otherwise be useless.) C-r <RET> does likewise, invoking the command search-backward.

Word search

A word search finds a sequence of words without regard to the type of punctuation between them. For instance, if you enter a search string that consists of two words separated by a single space, the search matches any sequence of those two words separated by one or more spaces, newlines, or other punctuation characters. This is particularly useful for searching text documents, because you don't have to worry whether the words you are looking for are separated by newlines or spaces.

To begin a forward incremental word search, type M-s w. If incremental search is not already active, this runs the command isearch-forward-word. If incremental search is already active (whether a forward or backward search), M-s w switches to a word search while keeping the direction of the search and the current search string unchanged. You can toggle word search back off by typing M-s w again.

To begin a nonincremental word search, type M-s w <RET> for a forward search, or M-s w C-r <RET> for a backward search. These run the commands word-search-forward and word-search-backward respectively.

Incremental and nonincremental word searches differ slightly in the way they find a match. In a nonincremental word search, the last word in the search string must exactly match a whole word. In an incremental word search, the matching is more lax: the last word in the search string can match part of a word, so that the matching proceeds incrementally as you type. This additional laxity does not apply to the lazy highlight, which always matches whole words.

Symbol search

A symbol search is much like an ordinary search, except that the boundaries of the search must match the boundaries of a symbol. The meaning of symbol in this context depends on the major mode, and usually refers to a source code token, such as a Lisp symbol in Emacs Lisp mode. For instance, if you perform an incremental symbol search for the Lisp symbol forward-word, it would not match isearch-forward-word. This feature is thus mainly useful for searching source code.

To begin a forward incremental symbol search, type M-s _. If incremental search is not already active, this runs the command isearch-forward-symbol. If incremental search is already active, M-s _ switches to a symbol search, preserving the direction of the search and the current search string; you can disable symbol search by typing M-s _ again. In incremental symbol search, only the beginning of the search string is required to match the beginning of a symbol.

To begin a nonincremental symbol search, type M-s _ <RET> for a forward search, or M-s _ C-r <RET> or a backward search. In nonincremental symbol searches, the beginning and end of the search string are required to match the beginning and end of a symbol, respectively.

Regular expression search

A regular expression (or regexp for short) is a pattern that denotes a class of alternative strings to match. Emacs provides both incremental and nonincremental ways to search for a match for a regexp. The syntax of regular expressions is explained in the next section.

Incremental search for a regexp is done by typing C-M-s (isearch-forward-regexp), by invoking C-s with a prefix argument (whose value does not matter), or by typing M-r within a forward incremental search. This command reads a search string incrementally only like C-s, but it treats the search string as a regexp rather than looking for an exact match against the text in the buffer. Each time you add text to the search string, you make the regexp longer, and the new regexp is searched. To search backward for a regexp, use C-M-r (isearch-backward-regexp), C-r with a prefix argument, or M-r within a backward incremental search.

All of the special key sequences in an ordinary incremental search do similar things in an incremental regexp search. For instance, typing C-s immediately after starting the search retrieves the last incremental search regexp used and searches forward for it. Incremental regexp and non-regexp searches have independent defaults. They also have separate search rings, which you can access with M-p and M-n.

As in ordinary incremental search, any <SPC> typed in incremental regexp search matches any sequence of one or more whitespace characters. The variable search-whitespace-regexp specifies the regexp for the lax space matching, and M-s <SPC> (isearch-toggle-lax-whitespace) toggles the feature. See Special Isearch.

In some cases, adding characters to the regexp in an incremental regexp search can make the cursor move back and start again. For example, if you have searched for 'foo' and you add '\|bar', the cursor backs up in case the first 'bar' precedes the first 'foo'. See Regexps.

Forward and backward regexp search are not symmetrical, because regexp matching in Emacs always operates forward, starting with the beginning of the regexp. Thus, forward regexp search scans forward, trying a forward match at each possible starting position. Backward regexp search scans backward, trying a forward match at each possible starting position. These search methods are not mirror images.

Nonincremental search for a regexp is done with the commands re-search-forward and re-search-backward. You can invoke these with M-x, or by way of incremental regexp search with C-M-s <RET> and C-M-r <RET>.

If you use the incremental regexp search commands with a prefix argument, they perform ordinary string search, like isearch-forward and isearch-backward. See Incremental Search.

Syntax of regular expressions

This manual describes regular expression features that users often use.

Regular expressions have a syntax where a few characters are special constructs and the rest are ordinary. An ordinary character matches that same character and nothing else. The special characters are '$^.*+?[\'. The character ']' is special if it ends a character alternative. The character '-' is special inside a character alternative. Any other character appearing in a regular expression is ordinary, unless a '\' precedes it. (When you use regular expressions in a Lisp program, each '\' must be doubled, see the example near the end of this section.)

For example, 'f' is not a special character, so it is ordinary, and therefore 'f' is a regular expression that matches the string 'f' and no other string. (It does not match the string 'ff'.) Likewise, 'o' is a regular expression that matches only 'o'. (When case distinctions are being ignored, these regexps also match 'F' and 'O', but we consider this a generalization of "the same string", rather than an exception.)

Any two regular expressions a and b can be concatenated. The result is a regular expression which matches a string if a matches some amount of the beginning of that string and b matches the rest of the string. For example, concatenating the regular expressions 'f' and 'o' gives the regular expression 'fo', which matches only the string 'fo'. Still, trivial. To do something nontrivial, you need to use one of the special characters. Here is a list of them:

Note: for historical compatibility, special characters are treated as ordinary ones if they are in contexts where their special meanings make no sense. For example, '*foo' treats '*' as ordinary since there is no preceding expression on which the '*' can act. It is poor practice to depend on this behavior; it is better to quote the special character, regardless of where it appears.

As a '\' is not special inside a character alternative, it can never remove the special meaning of '-' or ']'. So don't quote these characters when they have no special meaning either. This would not clarify anything, as backslashes can legitimately precede these characters where they have special meaning, as in '[^\]' ("[^\\]" for Lisp string syntax), which matches any single character except a backslash.

Backslash in regular expressions

For the most part, '\' followed by any character matches only that character. However, there's several exceptions: two-character sequences starting with '\' that have special meanings. The second character in the sequence is always an ordinary character when used on its own. Here is a table of '\' constructs.

The constructs that pertain to words and syntax are controlled by the setting of the syntax table.

Regular expression example

Here is an example of a regexp: similar to the regexp that Emacs uses, by default, to recognize the end of a sentence, not including the following space (i.e., the variable sentence-end-base):

[.?!][]\"')}]*

This contains two parts in succession: a character set matching period, '?', or '!', and a character set matching close-brackets, quotes, or parentheses, repeated zero or more times.

Searching and case

Searches in Emacs normally ignore the case of the text they are searching through, if you specify the text in lowercase. Thus, if you specify searching for 'foo', then 'Foo' and 'foo' also match. Regexps, and in particular character sets, behave likewise: '[ab]' matches 'a' or 'A' or 'b' or 'B'.

An upper-case letter anywhere in the incremental search string makes the search case-sensitive. Thus, searching for 'Foo' does not find 'foo' or 'FOO'. This applies to regular expression search and to string search. The effect ceases if you delete the upper-case letter from the search string.

Typing M-c within an incremental search toggles the case sensitivity of that search. The effect does not extend beyond the current incremental search to the next one, but it does override the effect of adding or removing an upper-case letter in the current search.

If you set the variable case-fold-search to nil, then all letters must match exactly, including case. This is a per-buffer variable; altering the variable normally affects only the current buffer, unless you change its default value. See Locals. This variable applies to nonincremental searches also, including those performed by the replace commands (see Replace) and the minibuffer history matching commands (see Minibuffer History).

Several related variables control case-sensitivity of searching and matching for specific commands or activities. For instance, tags-case-fold-search controls case sensitivity for find-tag. To find these variables, do M-x apropos-variable <RET> case-fold-search <RET>.

Replacement commands

Emacs provides several commands for performing search-and-replace operations. In addition to the simple M-x replace-string command, there is M-% (query-replace), which presents each occurrence of the pattern and asks you whether to replace it.

The replace commands normally operate on the text from point to the end of the buffer. When the region is active, they operate on it instead (see Mark). The basic replace commands replace one search string (or regexp) with one replacement string. It is possible to perform several replacements in parallel, using the command expand-region-abbrevs (see Expanding Abbrevs).

Unlike incremental search, the replacement commands do not use lax space matching (see Special Isearch) by default. To enable lax space matching for replacement, change the variable replace-lax-whitespace to t. (This only affects how Emacs finds the text to replace, not the replacement text.)

• Unconditional Replace: Replacing all matches for a string.
• Regexp Replace: Replacing all matches for a regexp.
• Replacement and Case: How replacements preserve case of letters.
• Query Replace: How to use querying.

Unconditional replacement

To replace every instance of 'foo' after point with 'bar', use the command M-x replace-string with the two arguments 'foo' and 'bar'. Replacement happens only in the text after point, so if you want to cover the whole buffer you must go to the beginning first. All occurrences up to the end of the buffer are replaced; to limit replacement to part of the buffer, activate the region around that part. When the region is active, replacement is limited to the region (see Mark).

When replace-string exits, it leaves point at the last occurrence replaced. It adds the prior position of point (where the replace-string command was issued) to the mark ring, without activating the mark; use C-u C-<SPC> to move back there. See Mark Ring.

A prefix argument restricts replacement to matches that are surrounded by word boundaries.

See Replacement and Case, for details about case-sensitivity in replace commands.

Regexp replacement

The M-x replace-string command replaces exact matches for a single string. The similar command M-x replace-regexp replaces any match for a specified pattern.

In replace-regexp, the newstring need not be constant: it can refer to all or part of what is matched by the regexp. '\&' in newstring stands for the entire match being replaced. '\d' in newstring, where d is a digit, stands for whatever matched the dth parenthesized grouping in regexp. (This is called a "back reference".) '\#' refers to the count of replacements already made in this command, as a decimal number. In the first replacement, '\#' stands for '0'; in the second, for '1'; etc. For example,

M-x replace-regexp <RET> c[ad]+r <RET> \&-safe <RET>

Replaces (for example) 'cadr' with 'cadr-safe' and 'cddr' with 'cddr-safe'.

M-x replace-regexp <RET> \(c[ad]+r\)-safe <RET> \1 <RET>

Performs the inverse transformation. To include a '\' in the text to replace with, you must enter '\\'.

To enter part of the replacement string by hand each time, use '\?' in the replacement string. Each replacement asks you to edit the replacement string in the minibuffer, putting point where the '\?' was.

The remainder of this subsection is intended for specialized tasks and requires knowledge of Lisp. Most readers can skip it.

You can use Lisp expressions to calculate parts of the replacement string. To do this, write '\,' followed by the expression in the replacement string. Each replacement calculates the value of the expression and converts it to text without quoting (if it's a string, this means using the string's contents), and uses it in the replacement string in place of the expression itself. If the expression is a symbol, one space in the replacement string after the symbol name goes with the symbol name, so the value replaces them both.

Inside such an expression, you can use some special sequences. '\&' and '\n' refer here, as usual, to the entire match as a string, and to a submatch as a string. '\n' may be multiple digits, and the value of '\n' is nil if subexpression n did not match. You can also use '\#&' and '\#n' to refer to those matches as numbers (this is valid when the match or submatch has the form of a numeral). '\#' here too stands for the number of already-completed replacements.

Repeating our example to exchange 'x' and 'y', we can thus do it also this way:

M-x replace-regexp <RET> \(x\)\|y <RET>

\,(if \1 "y" "x") <RET>

For computing replacement strings for '\,', the format function is often useful. For example, to add consecutively numbered strings like 'ABC00042' to columns 73 to 80 (unless they are already occupied), you can use

M-x replace-regexp <RET> ^.\{0,72\}$ <RET>

\,(format "%-72sABC%05d" \& \#) <RET>

Replace commands and case

If the first argument of a replace command is all lowercase, the command ignores case while searching for occurrences to replace—provided case-fold-search is non-nil. If case-fold-search is set to nil, case is always significant in all searches.

Also, when the newstring argument is all or partly lowercase, replacement commands try to preserve the case pattern of each occurrence. Thus, the command

M-x replace-string <RET> foo <RET> bar <RET>

Replaces a lowercase 'foo' with a lowercase 'bar', an all-caps 'FOO' with 'BAR', and a capitalized 'Foo' with 'Bar'. (These three alternatives -- lowercase, all caps, and capitalized -- are the only ones that replace-string can distinguish.)

If upper-case letters are used in the replacement string, they remain uppercase every time that text is inserted. If upper-case letters are used in the first argument, the second argument is always substituted exactly as given, with no case conversion. Likewise, if either case-replace or case-fold-search is set to nil, replacement is done without case conversion.

Query replace

To change only some of the occurrences of 'foo' to 'bar', not all of them, use M-% (query-replace). This command finds occurrences of 'foo' one by one, displays each occurrence and asks you whether to replace it. Aside from querying, query-replace works only like replace-string (see Unconditional Replace). In particular, it preserves case provided case-replace is non-nil, as it normally is (see Replacement and Case). A numeric argument means to consider only occurrences that are bounded by word-delimiter characters.

C-M-% performs regexp search and replace (query-replace-regexp). It works like replace-regexp except that it queries like query-replace.

These commands highlight the current match using the face query-replace. They highlight other matches using lazy-highlight only like incremental search (see Incremental Search). By default, query-replace-regexp shows the substituted replacement string for the current match in the minibuffer. To keep special sequences '\&' and '\n' unexpanded, customize query-replace-show-replacement variable.

The characters you can type when you are shown a match for the string or regexp are:

Some other characters are aliases for the ones listed above: y, n and q are equivalent to <SPC>, <DEL> and <RET>.

Aside from this, any other character exits the query-replace, and is then reread as part of a key sequence. Thus, if you type C-k, it exits the query-replace and then kills to end of line.

To restart a query-replace once it is exited, use C-x <ESC> <ESC>, which repeats the query-replace because it used the minibuffer to read its arguments. See C-x <ESC> <ESC>.

See Operating on Files, for the Dired Q command which performs query replace on selected files. See also Transforming File Names, for Dired commands to rename, copy, or link files by replacing regexp matches in file names.

Other search-and-loop commands

Here are some other commands that find matches for a regular expression. They all ignore case in matching, if the pattern contains no upper-case letters and case-fold-search is non-nil. Aside from occur and its variants, all operate on the text from point to the end of the buffer, or on the region if it's active.

Commands for fixing typos

In this chapter we describe commands that are useful when you catch a mistake while editing. The most fundamental of these commands is the undo command C-/ (also bound to C-x u and C-_). This undoes a single command, or a part of a command (as in the case of query-replace), or several consecutive character insertions. Consecutive repetitions of C-/ undo earlier and earlier changes, back to the limit of the undo information available.

Aside from the commands described here, you can erase text using deletion commands such as <DEL> (delete-backward-char). These were described earlier in this manual. See Erasing.

• Undo: The Undo commands.
• Transpose: Exchanging two characters, words, lines, lists...
• Fixing Case: Correcting case of last word entered.
• Spelling: Apply spelling checker to a word, or a whole file.

Undo

The undo command reverses recent changes in the buffer's text. Each buffer records changes individually, and the undo command always applies to the current buffer. You can undo all the changes in a buffer for as far back as the buffer's records go. Usually, each editing command makes a separate entry in the undo records, but some commands such as query-replace divide their changes into multiple entries for flexibility in undoing. Consecutive character insertion commands are usually grouped together into a single undo record, to make undoing less tedious.

To begin to undo, type C-/ (or its aliases, C-_ or C-x u). This undoes the most recent change in the buffer, and moves point back to where it was before that change. Consecutive repetitions of C-/ (or its aliases) undo earlier and earlier changes in the current buffer. If all the recorded changes have already been undone, the undo command signals an error.

Any command other than an undo command breaks the sequence of undo commands. Starting from that moment, the entire sequence of undo commands that you have only performed are themselves placed into the undo record, as a single set of changes. Therefore, to re-apply changes you have undone, type C-f or any other command that harmlessly breaks the sequence of undoing; then type C-/ to undo the undo command.

Alternatively, if you want to resume undoing, without redoing previous undo commands, use M-x undo-only. This is like undo, but will not redo changes you have only undone.

If you notice that a buffer was modified accidentally, the easiest way to recover is to type C-/ repeatedly until the stars disappear from the front of the mode line (see Mode Line). Whenever an undo command makes the stars disappear from the mode line, it indicates the buffer contents are the same as they were when the file was last read in or saved. If you do not remember whether you changed the buffer deliberately, type C-/ once. When you see the last change you made undone, you will see whether it was an intentional change. If it was an accident, leave it undone. If it was deliberate, redo the change as described above.

When there is an active region, any use of undo performs selective undo: it undoes the most recent change in the region, instead of the entire buffer. However, when Transient Mark mode is off (see Disabled Transient Mark), C-/ always operates on the entire buffer, ignoring the region. In this case, you can perform selective undo by supplying a prefix argument to the undo command: C-u C-/. To undo further changes in the same region, repeat the undo command (no prefix argument is needed).

Some specialized buffers do not make undo records. Buffers whose names start with spaces never do; these buffers are used internally by Emacs to hold text that users don't normally look at or edit.

When the undo information for a buffer becomes too large, Emacs discards the oldest records from time to time (during garbage collection). You can specify how much undo information to keep by setting the variables undo-limit, undo-strong-limit, and undo-outer-limit. Their values are expressed in bytes.

The variable undo-limit sets a soft limit: Emacs keeps undo data for enough commands to reach this size, and perhaps exceed it, but does not keep data for any earlier commands beyond that. Its default value is 80000. The variable undo-strong-limit sets a stricter limit: any previous command (though not the most recent one) that pushes the size past this amount is forgotten. The default value of undo-strong-limit is 120000. Regardless of the values of those variables, the most recent change is never discarded unless it gets bigger than undo-outer-limit (normally 12,000,000). At that point, Emacs discards the undo data and warns you about it. This is the only situation where you cannot undo the last command. If this happens, you can increase the value of undo-outer-limit to make it even less likely to happen in the future. But if you didn't expect the command to create such large undo data, then it is probably a bug.

Transposing text

The common error of transposing two characters can be fixed, when they are adjacent, with the C-t command (transpose-chars). Normally, C-t transposes the two characters on either side of point. When given at the end of a line, rather than transposing the last character of the line with the newline, which would be useless, C-t transposes the last two characters on the line. So, if you catch your transposition error right away, you can fix it with only a C-t. If you don't catch it so fast, you must move the cursor back between the two transposed characters before you type C-t. If you transposed a space with the last character of the word before it, the word motion commands are a good way of getting there. Otherwise, a reverse search (C-r) is often the best way. See Search.

M-t transposes the word before point with the word after point (transpose-words). It moves point forward over a word, dragging the word preceding or containing point forward as well. The punctuation characters between the words do not move. For example, 'FOO, BAR' transposes into 'BAR, FOO' rather than 'BAR FOO,'.

C-M-t (transpose-sexps) is a similar command for transposing two expressions (see Expressions), and C-x C-t (transpose-lines) exchanges lines. They work like M-t except as regards what units of text they transpose.

A numeric argument to a transpose command serves as a repeat count: it tells the transpose command to move the character (word, expression, line) before or containing point across other characters (words, expressions, lines). For example, C-u 3 C-t moves the character before point forward across three other characters. It would change 'f-!-oobar' into 'oobf-!-ar'. This is equivalent to repeating C-t three times. C-u - 4 M-t moves the word before point backward across four words. C-u - C-M-t would cancel the effect of plain C-M-t.

A numeric argument of zero is assigned a special meaning (because otherwise a command with a repeat count of zero would do nothing): to transpose the character (word, expression, line) ending after point with the one ending after the mark.

Case conversion

A very common error is to type words in the wrong case. Because of this, the word case-conversion commands M-l, M-u and M-c have a special feature when used with a negative argument: they do not move the cursor. As soon as you see you have mistyped the last word, you can case-convert it and go on typing. See Case.

Checking and correcting spelling

This section describes the commands to check the spelling of a single word or of a portion of a buffer. These commands only work if the spelling checker program Aspell, Ispell or Hunspell is installed. These programs are not part of Emacs, but one of them is usually installed in GNU/Linux and other free operating systems. See Aspell.

To check the spelling of the word around or before point, and optionally correct it as well, type M-$ (ispell-word). If a region is active, M-$ checks the spelling of all words in the region. See Mark. (When Transient Mark mode is off, M-$ always acts on the word around or before point, ignoring the region; see Disabled Transient Mark.)

Similarly, the command M-x ispell performs spell-checking in the region if one is active, or in the entire buffer otherwise. The commands M-x ispell-buffer and M-x ispell-region explicitly perform spell-checking on the entire buffer or the region respectively. To check spelling in an e-mail message you are writing, use M-x ispell-message; that command checks the whole buffer, except for material that is indented or appears to be cited from other messages. See Sending Mail.

When one of these commands encounters what appears to be an incorrect word, it asks you what to do. It usually displays a list of numbered "near-misses"—words that are close to the incorrect word. Then you must type a single-character response. Here are the valid responses:

In Text mode and related modes, M-<TAB> (ispell-complete-word) performs in-buffer completion based on spelling correction. Insert the beginning of a word, and then type M-<TAB>; this shows a list of completions. (If your window manager intercepts M-<TAB>, type <ESC> <TAB> or C-M-i.) Each completion is listed with a digit or character; type that digit or character to choose it.

Once started, the Aspell or Ispell or Hunspell subprocess continues to run, waiting for something to do, so that subsequent spell checking commands complete more quickly. To get rid of the process, use M-x ispell-kill-ispell. This is not usually necessary, as the process uses no processor time except when you do spelling correction.

Ispell, Aspell and Hunspell look up spelling in two dictionaries: the standard dictionary and your dictionary. The standard dictionary is specified by the variable ispell-local-dictionary or, if that is nil, by the variable ispell-dictionary. If both are nil, the spelling program's default dictionary is used. The command M-x ispell-change-dictionary sets the standard dictionary for the buffer and then restarts the subprocess, so that it uses a different standard dictionary. Your personal dictionary is specified by the variable ispell-personal-dictionary. If that is nil, the spelling program looks for a personal dictionary in a default location.

A separate dictionary is used for word completion. The variable ispell-complete-word-dict specifies the file name of this dictionary. The completion dictionary must be different because it cannot use root and affix information. For some languages, there is a spell checking dictionary but no word completion dictionary.

Flyspell mode is a minor mode that performs automatic spell checking as you type. When it finds a word that it does not recognize, it highlights that word. Type M-x flyspell-mode to toggle Flyspell mode in the current buffer. To enable Flyspell mode in all text mode buffers, add flyspell-mode to text-mode-hook. See Hooks.

When Flyspell mode highlights a word as misspelled, you can click it with Mouse-2 to display a menu of possible corrections and actions. You can also correct the word by editing it manually in any way you like.

Flyspell Prog mode works only like ordinary Flyspell mode, except that it only checks words in comments and string constants. This feature is useful for editing programs. Type M-x flyspell-prog-mode to enable or disable this mode in the current buffer. To enable this mode in all programming mode buffers, add flyspell-prog-mode to prog-mode-hook (see Hooks).

Keyboard macros

In this chapter we describe how to record a sequence of editing commands so you can repeat it conveniently later.

A keyboard macro is a command defined by an Emacs user to stand for another sequence of keys. For example, if you discover that you are about to type C-n M-d C-d forty times, you can speed your work by defining a keyboard macro to do C-n M-d C-d, and then executing it 39 more times.

You define a keyboard macro by executing and recording the commands that are its definition. Put differently, as you define a keyboard macro, the definition is being executed for the first time. This way, you can see the effects of your commands, so that you don't have to figure them out in your head. When you close the definition, the keyboard macro is defined and also is, in effect, executed once. You can then do the whole thing over again by invoking the macro.

Keyboard macros differ from ordinary Emacs commands in that they are written in the Emacs command language rather than in Lisp. This makes it easier for the novice to write them, and makes them more convenient as temporary hacks. However, the Emacs command language is not powerful enough as a programming language to be useful for writing anything intelligent or general. For such things, Lisp must be used.

• Basic Keyboard Macro: Defining and running keyboard macros.
• Keyboard Macro Ring: Where previous keyboard macros are saved.
• Keyboard Macro Counter: Inserting incrementing numbers in macros.
• Keyboard Macro Query: Making keyboard macros do different things each time.
• Save Keyboard Macro: Giving keyboard macros names; saving them in files.
• Edit Keyboard Macro: Editing keyboard macros.
• Keyboard Macro Step-Edit: Interactively executing and editing a keyboard macro.

Basic macro use

To start defining a keyboard macro, type <F3>. From then on, your keys continue to be executed, but also become part of the definition of the macro. 'Def' appears in the mode line to remind you of what is happening. When you are finished, type <F4> (kmacro-end-or-call-macro) to terminate the definition. For example,

<F3> M-f foo <F4>

Defines a macro to move forward a word and then insert 'foo'. Note that <F3> and <F4> do not become part of the macro.

After defining the macro, you can call it with <F4>. For the above example, this has the same effect as typing M-f foo again. (Note the two roles of the <F4> command: it ends the macro if you are defining one, or calls the last macro otherwise.) You can also supply <F4> with a numeric prefix argument 'n', which means to invoke the macro 'n' times. An argument of zero repeats the macro indefinitely, until it gets an error or you type C-g (or, on MS-DOS, C-<BREAK>).

The above example demonstrates a handy trick that you can employ with keyboard macros: if you want to repeat an operation at regularly spaced places in the text, include a motion command as part of the macro. In this case, repeating the macro inserts the string 'foo' after each successive word.

After terminating the definition of a keyboard macro, you can append more keystrokes to its definition by typing C-u <F3>. This is equivalent to plain <F3> followed by retyping the whole definition so far. As a consequence, it re-executes the macro as previously defined. If you change the variable kmacro-execute-before-append to nil, the existing macro is not re-executed before appending to it (the default is t). You can also add to the end of the definition of the last keyboard macro without re-executing it by typing C-u C-u <F3>.

When a command reads an argument with the minibuffer, your minibuffer input becomes part of the macro along with the command. So when you replay the macro, the command gets the same argument as when you entered the macro. For example,

<F3> C-a C-k C-x b foo <RET> C-y C-x b <RET> <F4>

Defines a macro that kills the current line, yanks it into the buffer 'foo', then returns to the original buffer.

Most keyboard commands work as usual in a keyboard macro definition, with some exceptions. Typing C-g (keyboard-quit) quits the keyboard macro definition. Typing C-M-c (exit-recursive-edit) can be unreliable: it works as you'd expect if exiting a recursive edit that started in the macro, but if it exits a recursive edit that started before you invoked the keyboard macro, it also necessarily exits the keyboard macro too. Mouse events are also unreliable, even though you can use them in a keyboard macro: when the macro replays the mouse event, it uses the original mouse position of that event, the position that the mouse had while you were defining the macro. The effect of this may be hard to predict.

The command C-x C-k r (apply-macro-to-region-lines) repeats the last defined keyboard macro on each line that begins in the region. It does this line by line, by moving point to the beginning of the line and then executing the macro.

In addition to the <F3> and <F4> commands described above, Emacs also supports an older set of key bindings for defining and executing keyboard macros. To begin a macro definition, type C-x ( (kmacro-start-macro); as with <F3>, a prefix argument appends this definition to the last keyboard macro. To end a macro definition, type C-x ) (kmacro-end-macro). To execute the most recent macro, type C-x e (kmacro-end-and-call-macro). If you enter C-x e while defining a macro, the macro is terminated and executed immediately. Immediately after typing C-x e, you can type <e> repeatedly to immediately repeat the macro one or more times. You can also give C-x e a repeat argument, only like <F4>.

C-x ) can be given a repeat count as an argument. This means to repeat the macro right after defining it. The macro definition itself counts as the first repetition, as it is executed as you define it, so C-u 4 C-x ) executes the macro immediately 3 additional times.

The keyboard macro ring

All defined keyboard macros are recorded in the keyboard macro ring. There is only one keyboard macro ring, shared by all buffers.

All commands which operate on the keyboard macro ring use the same C-x C-k prefix. Most of these commands can be executed and repeated immediately after each other without repeating the C-x C-k prefix. For example,

C-x C-k C-p C-p C-k C-k C-k C-n C-n C-k C-p C-k C-d

will rotate the keyboard macro ring to the "second previous" macro, execute the resulting head macro three times, rotate back to the original head macro, execute that once, rotate to the "previous" macro, execute that, and finally delete it from the macro ring.

The command C-x C-k C-k (kmacro-end-or-call-macro-repeat) executes the keyboard macro at the head of the macro ring. You can repeat the macro immediately by typing another C-k, or you can rotate the macro ring immediately by typing C-n or C-p.

When a keyboard macro is being defined, C-x C-k C-k behaves like <F4> except that, immediately afterward, you can use most key bindings of this section without the C-x C-k prefix. For instance, another C-k will re-execute the macro.

The commands C-x C-k C-n (kmacro-cycle-ring-next) and C-x C-k C-p (kmacro-cycle-ring-previous) rotate the macro ring, bringing the next or previous keyboard macro to the head of the macro ring. The definition of the new head macro is displayed in the echo area. You can continue to rotate the macro ring immediately by repeating only C-n and C-p until the desired macro is at the head of the ring. To execute the new macro ring head immediately, only type C-k.

Note that Emacs treats the head of the macro ring as the "last defined keyboard macro". For instance, <F4> will execute that macro, and C-x C-k n gives it a name.

The maximum number of macros stored in the keyboard macro ring is determined by the customizable variable kmacro-ring-max.

The keyboard macro counter

Each keyboard macro has an associated counter, which is initialized to 0 when you start defining the macro. This counter allows you to insert a number into the buffer that depends on the number of times the macro is called. The counter is incremented each time its value is inserted into the buffer.

When you are defining a keyboard macro, the command <F3> (kmacro-start-macro-or-insert-counter) inserts the current value of the keyboard macro's counter into the buffer, and increments the counter by 1. (If you are not defining a macro, <F3> begins a macro definition instead. See Basic Keyboard Macro.) You can use a numeric prefix argument to specify a different increment. If you only specify a C-u prefix, that is the same as an increment of zero: it inserts the current counter value without changing it.

As an example, let us show how the keyboard macro counter can build a numbered list. Consider the following key sequence:

<F3> C-a <F3> . <SPC> <F4>

As part of this keyboard macro definition, the string '0. ' was inserted into the beginning of the current line. If you now move somewhere else in the buffer and type <F4> to invoke the macro, the string '1. ' is inserted at the beginning of that line. Subsequent invocations insert '2. ', '3. ', and so forth.

The command C-x C-k C-i (kmacro-insert-counter) does the same thing as <F3>, but it can be used outside a keyboard macro definition. When no keyboard macro is being defined or executed, it inserts and increments the counter of the macro at the head of the keyboard macro ring.

The command C-x C-k C-c (kmacro-set-counter) sets the current macro counter to the value of the numeric argument. If you use it inside the macro, it operates on each repetition of the macro. If you specify only C-u as the prefix, while executing the macro, that resets the counter to the value it had at the beginning of the current repetition of the macro (undoing any increments so far in this repetition).

The command C-x C-k C-a (kmacro-add-counter) adds the prefix argument to the current macro counter. With only C-u as argument, it resets the counter to the last value inserted by any keyboard macro. (Normally, when you use this, the last insertion is in the same macro and it is the same counter.)

The command C-x C-k C-f (kmacro-set-format) prompts for the format to use when inserting the macro counter. The default format is '%d', which means to insert the number in decimal without any padding. You can exit with empty minibuffer to reset the format to this default. You can specify any format string that the format function accepts and that makes sense with a single integer extra argument. Do not put the format string inside double quotes when you insert it in the minibuffer.

If you use this command while no keyboard macro is being defined or executed, the new format affects all subsequent macro definitions. Existing macros continue to use the format in effect when they were defined. If you set the format while defining a keyboard macro, this affects the macro being defined from that point on, but it does not affect subsequent macros. Execution of the macro, at each step, uses the format in effect at that step during its definition. Changes to the macro format during execution of a macro, like the corresponding changes during its definition, have no effect on subsequent macros.

The format set by C-x C-k C-f does not affect insertion of numbers stored in registers.

If you use a register as a counter, incrementing it on each repetition of the macro, that accomplishes the same thing as a keyboard macro counter. See Number Registers. For most purposes, it is simpler to use a keyboard macro counter.

Executing macros with variations

In a keyboard macro, you can create an effect similar to that of query-replace, in that the macro asks you each time around whether to make a change.

While defining the macro, type C-x q at the point where you want the query to occur. During macro definition, the C-x q does nothing, but when you run the macro later, C-x q asks you interactively whether to continue.

The valid responses when C-x q asks are:

C-u C-x q, which is C-x q with a numeric argument, performs a completely different function. It enters a recursive edit reading input from the keyboard, both when you type it during the definition of the macro, and when it is executed from the macro. During definition, the editing you do inside the recursive edit does not become part of the macro. During macro execution, the recursive edit gives you a chance to do some particularized editing on each repetition. See Recursive Edit.

Naming and saving keyboard macros

To save a keyboard macro for later use, you can give it a name using C-x C-k n (kmacro-name-last-macro). This reads a name as an argument using the minibuffer and defines that name to execute the last keyboard macro, in its current form. (If you later add to the definition of this macro, that does not alter the name's definition as a macro.) The macro name is a Lisp symbol, and defining it in this way makes it a valid command name for calling with M-x or for binding a key to with global-set-key (see Keymaps). If you specify a name with a prior definition other than a keyboard macro, an error message is shown and nothing is changed.

You can also bind the last keyboard macro (in its current form) to a key, using C-x C-k b (kmacro-bind-to-key) followed by the key sequence you want to bind. You can bind to any key sequence in the global keymap, but since most key sequences already have other bindings, select the key sequence carefully. If you try to bind to a key sequence with an existing binding (in any keymap), this command asks you for confirmation before replacing the existing binding.

To avoid problems caused by overriding existing bindings, the key sequences C-x C-k 0 through C-x C-k 9 and C-x C-k A through C-x C-k Z are reserved for your keyboard macro bindings. In fact, to bind to one of these key sequences, you only need to type the digit or letter rather than the whole key sequences. For example,

C-x C-k b 4

Binds the last keyboard macro to the key sequence C-x C-k 4.

Once a macro has a command name, you can save its definition in a file. Then it can be used in another editing session. First, visit the file where you want to save the definition. Then use this command:

M-x insert-kbd-macro <RET> macroname <RET>

This inserts some Lisp code that, when executed later, defines the same macro with the same definition it has now. (You need not understand Lisp code to do this, because insert-kbd-macro writes the Lisp code for you.) Then save the file. You can load the file later with load-file (see Lisp Libraries). If the file you save in is your init file ~/.emacs (see Init File) then the macro is defined each time you run Emacs.

If you give insert-kbd-macro a numeric argument, it makes additional Lisp code to record the keys (if any) that you have bound to macroname, so that the macro is reassigned the same keys when you load the file.

Editing a keyboard macro

You can edit the last keyboard macro by typing C-x C-k C-e or C-x C-k <RET> (kmacro-edit-macro). This formats the macro definition in a buffer and enters a specialized major mode for editing it. Type C-h m once in that buffer to display details of how to edit the macro. When you are finished editing, type C-c C-c.

You can edit a named keyboard macro or a macro bound to a key by typing C-x C-k e (edit-kbd-macro). Follow that with the keyboard input that you would use to invoke the macro: C-x e or M-x name or some other key sequence.

You can edit the last 300 keystrokes as a macro by typing C-x C-k l (kmacro-edit-lossage).

Stepwise editing a keyboard macro

You can interactively replay and edit the last keyboard macro, one command at a time, by typing C-x C-k <SPC> (kmacro-step-edit-macro). Unless you quit the macro using q or C-g, the edited macro replaces the last macro on the macro ring.

This macro editing feature shows the last macro in the minibuffer together with the first (or next) command to be executed, and prompts you for an action. You can enter ? to get a summary of your options. These actions are available:

• <SPC> and y execute the current command, and advance to the next command in the keyboard macro.
• n, d, and <DEL> skip and delete the current command.
• f skips the current command in this execution of the keyboard macro, but doesn't delete it from the macro.
• <TAB> executes the current command, and all similar commands immediately following the current command; for example, <TAB> may be used to insert a sequence of characters (corresponding to a sequence of self-insert-command commands).
• c continues execution (without further editing) until the end of the keyboard macro. If execution terminates normally, the edited macro replaces the original keyboard macro.
• C-k skips and deletes the rest of the keyboard macro, terminates step-editing, and replaces the original keyboard macro with the edited macro.
• q and C-g cancels the step-editing of the keyboard macro; discarding any changes made to the keyboard macro.
• i KEY... C-j reads and executes a series of key sequences (not including the final C-j), and inserts them before the current command in the keyboard macro, without advancing over the current command.
• I KEY... reads one key sequence, executes it, and inserts it before the current command in the keyboard macro, without advancing over the current command.
• r KEY... C-j reads and executes a series of key sequences (not including the final C-j), and replaces the current command in the keyboard macro with them, advancing over the inserted key sequences.
• R KEY... reads one key sequence, executes it, and replaces the current command in the keyboard macro with that key sequence, advancing over the inserted key sequence.
• a KEY... C-j executes the current command, then reads and executes a series of key sequences (not including the final C-j), and inserts them after the current command in the keyboard macro; it then advances over the current command and the inserted key sequences.
• A KEY... C-j executes the rest of the commands in the keyboard macro, then reads and executes a series of key sequences (not including the final C-j), and appends them at the end of the keyboard macro; it then terminates the step-editing and replaces the original keyboard macro with the edited macro.

File handling

The operating system stores data permanently in named files, so most of the text you edit with Emacs comes from a file and is ultimately stored in a file.

To edit a file, you must tell Emacs to read the file and prepare a buffer containing a copy of the file's text. This is called visiting the file. Editing commands apply directly to text in the buffer; that is, to the copy inside Emacs. Your changes appear in the file itself only when you save the buffer back into the file.

In addition to visiting and saving files, Emacs can delete, copy, rename, and append to files, keep multiple versions of them, and operate on file directories.

• File Names: How to type and edit file-name arguments.
• Visiting: Visiting a file prepares Emacs to edit the file.
• Saving: Saving makes your changes permanent.
• Reverting: Reverting cancels all the changes not saved.
• Autorevert: Auto Reverting non-file buffers.
• Auto Save: Auto Save periodically protects against loss of data.
• File Aliases: Handling multiple names for one file.
• Directories: Creating, deleting, and listing file directories.
• Comparing Files: Finding where two files differ.
• Diff Mode: Mode for editing file differences.
• Misc File Ops: Other things you can do on files.
• Compressed Files: Accessing compressed files.
• File Archives: Operating on tar, zip, jar etc. archive files.
• Remote Files: Accessing files on other machines.
• Quoted File Names: Quoting special characters in file names.
• File Name Cache: Completion against a list of files you often use.
• File Conveniences: Convenience Features for Finding Files.
• Filesets: Handling sets of files.

File names

Many Emacs commands that operate on a file require you to specify the file name, using the minibuffer (see Minibuffer File).

While in the minibuffer, you can use the usual completion and history commands (see Minibuffer). Note that file name completion ignores file names whose extensions appear in the variable completion-ignored-extensions (see Options). Note also that most commands use "permissive completion with confirmation" for reading file names: you are allowed to submit a nonexistent file name, but if you type <RET> immediately after completing up to a nonexistent file name, Emacs prints '[Confirm]' and you must type a second <RET> to confirm. See Completion Exit, for details.

Each buffer has a default directory, stored in the buffer-local variable default-directory. Whenever Emacs reads a file name using the minibuffer, it usually inserts the default directory into the minibuffer as the initial contents. You can inhibit this insertion by changing the variable insert-default-directory to nil (see Minibuffer File). Regardless, Emacs always assumes that any relative file name is relative to the default directory, e.g., entering a file name without a directory specifies a file in the default directory.

When you visit a file, Emacs sets default-directory in the visiting buffer to the directory of its file. When you create a new buffer that is not visiting a file, via a command like C-x b, its default directory is usually copied from the buffer that was current at the time (see Select Buffer). You can use the command M-x pwd to see the value of default-directory in the current buffer. The command M-x cd prompts for a directory name, and sets the buffer's default-directory to that directory (doing this does not change the buffer's file name, if any).

As an example, when you visit the file /u/rms/gnu/gnu.tasks, the default directory is set to /u/rms/gnu/. If you invoke a command that reads a file name, entering only 'foo' in the minibuffer, with a directory omitted, specifies the file /u/rms/gnu/foo; entering '../.login' specifies /u/rms/.login; and entering 'new/foo' specifies /u/rms/gnu/new/foo.

When typing a file name into the minibuffer, you can make use of a few shortcuts: a double slash is interpreted as "ignore everything before the second slash in the pair", and '~/' is interpreted as your home directory. See Minibuffer File.

The character '$' is used to substitute an environment variable into a file name. The name of the environment variable consists of all the alphanumeric characters after the '$'; alternatively, it can be enclosed in braces after the '$'. For example, if you have used the shell command export FOO=rms/hacks to set up an environment variable named FOO, then both /u/$FOO/test.c and /u/${FOO}/test.c are abbreviations for /u/rms/hacks/test.c. If the environment variable is not defined, no substitution occurs, so that the character '$' stands for itself. Note that environment variables affect Emacs only if they are applied before Emacs is started.

To access a file with '$' in its name, if the '$' causes expansion, type '$$'. This pair is converted to a single '$' at the same time that variable substitution is performed for a single '$'. Alternatively, quote the whole file name with '/:' (see Quoted File Names). File names which begin with a literal '~' should also be quoted with '/:'.

You can include non-ASCII characters in file names. See File Name Coding.

Visiting files

Visiting a file means reading its contents into an Emacs buffer so you can edit them. Emacs makes a new buffer for each file that you visit.

To visit a file, type C-x C-f (find-file) and use the minibuffer to enter the name of the desired file. While in the minibuffer, you can abort the command by typing C-g. See File Names, for details about entering file names into minibuffers.

If the specified file exists but the system does not allow you to read it, an error message is displayed in the echo area. Otherwise, you can tell that C-x C-f has completed successfully by the appearance of new text on the screen, and by the buffer name shown in the mode line (see Mode Line). Emacs normally constructs the buffer name from the file name, omitting the directory name. For example, a file named /usr/rms/emacs.tex is visited in a buffer named 'emacs.tex'. If there is already a buffer with that name, Emacs constructs a unique name; the normal method is to append '<2>', '<3>', and so on, but you can select other methods. See Uniquify.

To create a new file, only visit it using the same command, C-x C-f. Emacs displays '(New file)' in the echo area, but in other respects behaves as if you had visited an existing empty file.

After visiting a file, the changes you make with editing commands are made in the Emacs buffer. They do not take effect in the visited file, until you save the buffer (see Saving). If a buffer contains changes that have not been saved, we say the buffer is modified. This implies that some changes are lost if the buffer is not saved. The mode line displays two stars near the left margin to indicate that the buffer is modified.

If you visit a file that is already in Emacs, C-x C-f switches to the existing buffer instead of making another copy. Before doing so, it checks whether the file has changed since you last visited or saved it. If the file has changed, Emacs offers to reread it.

If you try to visit a file larger than large-file-warning-threshold (the default is 10000000, which is about 10 megabytes), Emacs asks you for confirmation first. You can answer y to proceed with visiting the file. Note, however, that Emacs cannot visit files that are larger than the maximum Emacs buffer size, which is limited by the amount of memory Emacs can allocate and by the integers that Emacs can represent (see Buffers). If you try, Emacs displays an error message saying that the maximum buffer size was exceeded.

If the file name you specify contains shell-style wildcard characters, Emacs visits all the files that match it. (On case-insensitive filesystems, Emacs matches the wildcards disregarding the letter case.) Wildcards include '?', '*', and '[...]' sequences. To enter the wildcard '?' in a file name in the minibuffer, you need to type C-q ?. See Quoted File Names, for information on how to visit a file whose name actually contains wildcard characters. You can disable the wildcard feature by customizing find-file-wildcards.

If you visit the wrong file unintentionally by typing its name incorrectly, type C-x C-v (find-alternate-file) to visit the file you wanted. C-x C-v is similar to C-x C-f, but it kills the current buffer (after first offering to save it if it's modified). When C-x C-v reads the file name to visit, it inserts the entire default file name in the buffer, with point only after the directory part; this is convenient if you made a slight error in typing the name.

If you "visit" a file that is actually a directory, Emacs invokes Dired, the Emacs directory browser. See Dired. You can disable this behavior by setting the variable find-file-run-dired to nil; in that case, it is an error to try to visit a directory.

Files that are actually collections of other files, or file archives, are visited in special modes which invoke a Dired-like environment to allow operations on archive members. See File Archives, for more about these features.

If you visit a file that the operating system won't let you modify, or that is marked read-only, Emacs makes the buffer read-only too, so that you won't go ahead and make changes that you'll have trouble saving afterward. You can make the buffer writable with C-x C-q (read-only-mode). See Misc Buffer.

To visit a file as read-only to protect yourself from entering changes accidentally, visit it with the command C-x C-r (find-file-read-only) instead of C-x C-f.

C-x 4 f (find-file-other-window) is like C-x C-f except that the buffer containing the specified file is selected in another window. The window that was selected before C-x 4 f continues to show the same buffer it was already showing. If this command is used when only one window is being displayed, that window is split in two, with one window showing the same buffer as before, and the other one showing the newly requested file. See Windows.

C-x 5 f (find-file-other-frame) is similar, but opens a new frame, or selects any existing frame showing the specified file. See Frames.

On graphical displays, there are two additional methods for visiting files. Firstly, when Emacs is built with a suitable GUI toolkit, commands invoked with the mouse (by clicking the menu bar or tool bar) use the toolkit's standard "File Selection" dialog instead of prompting for the file name in the minibuffer. On GNU/Linux platforms, Emacs does this when built with GTK, LessTif, and Motif toolkits; on Microsoft Windows and Mac, the GUI version does that by default. For information on how to customize this, see Dialog Boxes.

Secondly, Emacs supports "drag-and-drop": dropping a file into an ordinary Emacs window visits the file using that window. As an exception, dropping a file into a window displaying a Dired buffer moves or copies the file into the displayed directory. For details, see Drag and Drop, and Misc Dired Features.

Each time you visit a file, Emacs automatically scans its contents to detect what character encoding and end-of-line convention it uses, and converts these to Emacs's internal encoding and end-of-line convention in the buffer. When you save the buffer, Emacs performs the inverse conversion, writing the file to disk with its original encoding and end-of-line convention. See Coding Systems.

To edit a file as a sequence of ASCII characters with no special encoding or conversion, use the M-x find-file-literally command. This visits a file, like C-x C-f, but does not do format conversion, character code conversion (see Coding Systems), or automatic uncompression (see Compressed Files), and does not add a final newline because of require-final-newline (see Customize Save). If you have already visited the same file in the usual (non-literal) manner, this command asks you whether to visit it literally instead.

Two special hook variables allow extensions to modify the operation of visiting files. Visiting a file that does not exist runs the functions in find-file-not-found-functions; this variable holds a list of functions, which are called one by one (with no arguments) until one of them returns non-nil. This is not a normal hook, and the name ends in '-functions' rather than '-hook' to indicate that fact.

Successful visiting of any file, whether existing or not, calls the functions in find-file-hook, with no arguments. This variable is a normal hook. In the case of a nonexistent file, the find-file-not-found-functions are run first. See Hooks.

There's several ways to specify automatically the major mode for editing the file (see Choosing Modes), and to specify local variables defined for that file (see File Variables).

Saving files

Saving a buffer in Emacs means writing its contents back into the file that was visited in the buffer.

• Save Commands: Commands for saving files.
• Backup: How Emacs saves the old version of your file.
• Customize Save: Customizing the saving of files.
• Interlocking: How Emacs protects against simultaneous editing of one file by two users.
• Shadowing: Copying files to "shadows" automatically.
• Timestamps: Emacs can update timestamps on saved files.

Commands for saving files

These are the commands that relate to saving and writing files.

When you want to save the file and make your changes permanent, type C-x C-s (save-buffer). After saving is finished, C-x C-s displays a message like this:

Wrote /u/rms/gnu/gnu.tasks

If the current buffer is not modified (no changes are made in it since the buffer was created or last saved), saving is not really done, because it would have no effect. Instead, C-x C-s displays a message like this in the echo area:

(No changes need to be saved)

With a prefix argument, C-u C-x C-s, Emacs also marks the buffer to be backed up when the next save is done. See Backup.

The command C-x s (save-some-buffers) offers to save any or all modified buffers. It asks you what to do with each buffer. The possible responses are analogous to those of query-replace:

C-x C-c, the key sequence to exit Emacs, invokes save-some-buffers and therefore asks the same questions.

If you have changed a buffer but do not want to save the changes, take some action to prevent it. Otherwise, each time you use C-x s or C-x C-c, you are liable to save this buffer by mistake. One thing you can do is type M-~ (not-modified), which clears out the indication that the buffer is modified. If you do this, none of the save commands believe the buffer needs to be saved. ('~' is often used as a mathematical symbol for `not'; thus M-~ is `not', metafied.) Alternatively, you can cancel all the changes made since the file was visited or saved, by reading the text from the file again. This is called reverting. See Reverting. (You could also undo all the changes by repeating the undo command C-x u until you have undone all the changes; but reverting is easier.)

M-x set-visited-file-name alters the name of the file that the current buffer is visiting. It reads the new file name using the minibuffer. Then it marks the buffer as visiting that file name, and changes the buffer name correspondingly. set-visited-file-name does not save the buffer in the newly visited file; it only alters the records inside Emacs in case you do save later. It also marks the buffer as "modified" so that C-x C-s in that buffer saves.

To mark the buffer as visiting a different file and save it right away, use C-x C-w (write-file). This is equivalent to set-visited-file-name followed by C-x C-s, except that C-x C-w asks for confirmation if the file exists. C-x C-s used on a buffer that is not visiting a file has the same effect as C-x C-w; that is, it reads a file name, marks the buffer as visiting that file, and saves it there. The default file name in a buffer that is not visiting a file is made by combining the buffer name with the buffer's default directory (see File Names).

If the new file name implies a major mode, then C-x C-w switches to that major mode, in most cases. The command set-visited-file-name also does this. See Choosing Modes.

If Emacs is about to save a file and sees that the date of the latest version on disk does not match what Emacs last read or wrote, Emacs notifies you of this fact, because it probably indicates a problem caused by simultaneous editing and requires your immediate attention.

Backing up files

On most operating systems, rewriting a file automatically destroys all record of what the file used to contain. Thus, saving a file from Emacs throws away the old contents of the file—or it would, except that Emacs carefully copies the old contents to another file, called the backup file, before actually saving.

Emacs makes a backup for a file only the first time the file is saved from a buffer. No matter how many times you subsequently save the file, its backup remains unchanged. However, if you kill the buffer and then visit the file again, a new backup file is made.

For most files, the variable make-backup-files determines whether to make backup files. On most operating systems, its default value is t, so that Emacs does write backup files.

For files managed by a version control system (see Version Control), the variable vc-make-backup-files determines whether to make backup files. By default, it is nil, as backup files are redundant when you store all the previous versions in a version control system. See Options.

At your option, Emacs can keep either a single backup for each file, or make a series of numbered backup files for each file that you edit. See Backup Names.

The default value of the backup-enable-predicate variable prevents backup files being written for files in the directories used for temporary files, specified by temporary-file-directory or small-temporary-file-directory.

You can explicitly tell Emacs to make another backup file from a buffer, even though that buffer was saved before. If you save the buffer with C-u C-x C-s, the version thus saved is made into a backup file if you save the buffer again. C-u C-u C-x C-s saves the buffer, but first makes the previous file contents into a new backup file. C-u C-u C-u C-x C-s does both things: it makes a backup from the previous contents, and arranges to make another from the newly saved contents if you save again.

• Names: How backup files are named.
• Deletion: Emacs deletes excess numbered backups.
• Copying: Backups can be made by copying or renaming.

Single or numbered backups

When Emacs makes a backup file, its name is normally constructed by appending '~' to the file name being edited; thus, the backup file for eval.c would be eval.c~.

If access control stops Emacs from writing backup files under the usual names, it writes the backup file as ~/.emacs.d/%backup%~. Only one such file can exist, so only the most recently made such backup is available.

Emacs can also make numbered backup files. Numbered backup file names contain '.~', the number, and another '~' after the original file name. Thus, the backup files of eval.c would be called eval.c.~1~, eval.c.~2~, and so on, through names like eval.c.~259~ and beyond.

The variable version-control determines whether to make single backup files or multiple numbered backup files. Its possible values are:

The usual way to set this variable is globally, through your init file or the customization buffer. However, you can set version-control locally in an individual buffer to control the making of backups for that buffer's file (see Locals). You can have Emacs set version-control locally whenever you visit a given file (see File Variables). Some modes, such as Rmail mode, set this variable.

If you set the environment variable VERSION_CONTROL, to tell various GNU utilities what to do with backup files, Emacs also obeys the environment variable by setting the Lisp variable version-control accordingly at startup. If the environment variable's value is 't' or 'numbered', then version-control becomes t; if the value is 'nil' or 'existing', then version-control becomes nil; if it's 'never' or 'simple', then version-control becomes never.

You can customize the variable backup-directory-alist to specify that files matching certain patterns should be backed up in specific directories. This variable applies to both single and numbered backups. A typical use is to add an element ("." . dir) to make all backups in the directory with absolute name dir; Emacs modifies the backup file names to avoid clashes between files with the same names originating in different directories. Alternatively, adding, ("." . ".~") would make backups in the invisible subdirectory .~ of the original file's directory. Emacs creates the directory, if necessary, to make the backup.

If you define the variable make-backup-file-name-function to a suitable Lisp function, that overrides the usual way Emacs constructs backup file names.

Automatic deletion of backups

To prevent excessive consumption of disk space, Emacs can delete numbered backup versions automatically. Generally Emacs keeps the first few backups and the latest few backups, deleting any between. This happens every time a new backup is made.

The two variables kept-old-versions and kept-new-versions control this deletion. Their values are, respectively, the number of oldest (lowest-numbered) backups to keep and the number of newest (highest-numbered) ones to keep, each time a new backup is made. The backups in the middle (excluding those oldest and newest) are the excess middle versions: those backups are deleted. These variables' values are used when it is time to delete excess versions, only after a new backup version is made; the newly made backup is included in the count in kept-new-versions. By default, both variables are 2.

If delete-old-versions is t, Emacs deletes the excess backup files silently. If it's nil, the default, Emacs asks you whether it should delete the excess backup versions. If it has any other value, then Emacs never automatically deletes backups.

Dired's . (Period) command can also be used to delete old versions. See Dired Deletion.

Copying vs. renaming

Backup files can be made by copying the old file or by renaming it. This makes a difference when the old file has multiple names (hard links). If the old file is renamed into the backup file, then the alternate names become names for the backup file. If the old file is copied instead, then the alternate names remain names for the file that you are editing, and the contents accessed by those names are the new contents.

The method of making a backup file may also affect the file's owner and group. If copying is used, these do not change. If renaming is used, you become the file's owner, and the file's group becomes the default (different operating systems have different defaults for the group).

The choice of renaming or copying is made as follows:

• If the variable backup-by-copying is non-nil (the default is nil), use copying.
• Otherwise, if the variable backup-by-copying-when-linked is non-nil (the default is nil), and the file has multiple names, use copying.
• Otherwise, if the variable backup-by-copying-when-mismatch is non-nil (the default is t), and renaming would change the file's owner or group, use copying. If you change backup-by-copying-when-mismatch to nil, Emacs checks the numeric user-id of the file's owner. If this is higher than backup-by-copying-when-privileged-mismatch, then it behaves as though backup-by-copying-when-mismatch is non-nil.
• Otherwise, renaming is the default choice.

When a file is managed with a version control system (see Version Control), Emacs does not normally make backups in the usual way for that file. But check-in and check-out are similar in some ways to making backups. One unfortunate similarity is that these operations often break hard links, disconnecting the file name you visited from any alternate names for the same file. This has nothing to do with Emacs—the version control system does it.

Customizing the saving of files

If the value of the variable require-final-newline is t, saving or writing a file silently puts a newline at the end if there isn't already one there. If the value is visit, Emacs adds a newline at the end of any file that doesn't have one, only after it visits the file. (This marks the buffer as modified, and you can undo it.) If the value is visit-save, Emacs adds such newlines both on visiting and on saving. If the value is nil, Emacs leaves the end of the file unchanged; any other non-nil value means to asks you whether to add a newline. The default is nil.

Some major modes are designed for specific kinds of files that are always supposed to end in newlines. Such major modes set the variable require-final-newline to the value of mode-require-final-newline, which defaults to t. By setting the latter variable, you can control how these modes handle final newlines.

When Emacs saves a file, it invokes the fsync system call to force the data immediately out to disk. This is important for safety if the system crashes or in case of power outage. However, it can be disruptive on laptops using power saving, as it may force a disk spin-up each time you save a file. If you accept an increased risk of data loss, you can set write-region-inhibit-fsync to a non-nil value to disable the synchronization.

Protection against simultaneous editing

Simultaneous editing occurs when two users visit the same file, both make changes, and then both save them. If nobody is informed that this is happening, whichever user saves first would later find that his changes were lost.

On some systems, Emacs notices immediately when the second user starts to change the file, and issues an immediate warning. On all systems, Emacs checks when you save the file, and warns if you are about to overwrite another user's changes. You can prevent loss of the other user's work by taking the proper corrective action instead of saving the file.

When you make the first modification in an Emacs buffer that is visiting a file, Emacs records that the file is locked by you. (It does this by creating a specially-named symbolic link in the same directory.) Emacs removes the lock when you save the changes. The idea is that the file is locked whenever an Emacs buffer visiting it has unsaved changes.

You can prevent the creation of lock files by setting the variable create-lockfiles to nil. Caution: by doing so, you lose the benefits that this feature provides.

If you begin to modify the buffer while the visited file is locked by someone else, this constitutes a collision. When Emacs detects a collision, it asks you what to do, by calling the Lisp function ask-user-about-lock. You can redefine this function for the sake of customization. The standard definition of this function asks you a question and accepts three possible answers:

If Emacs or the operating system crashes, this may leave behind lock files that are stale, so you may occasionally get warnings about spurious collisions. When you determine that the collision is spurious, only use p to tell Emacs to go ahead.

Note that locking works on the basis of a file name; if a file has multiple names, Emacs does not prevent two users from editing it simultaneously under different names.

A lock file cannot be written in some circumstances, e.g., if Emacs lacks the system permissions or the system does not support symbolic links. In these cases, Emacs can still detect the collision when you try to save a file, by checking the file's last-modification date. If the file has changed since the last time Emacs visited or saved it, that implies that changes were made in some other way, and are lost if Emacs proceeds with saving. Emacs then displays a warning message and asks for confirmation before saving; answer yes to save, and no or C-g cancel the save.

If you are notified that simultaneous editing has already taken place, one way to compare the buffer to its file is the M-x diff-buffer-with-file command. See Comparing Files.

Shadowing files

You can arrange to keep identical shadow copies of certain files in more than one place—possibly on different machines. To do this, first you must set up a shadow file group, which is a set of identically-named files shared between a list of sites. The file group is permanent and applies to further Emacs sessions and the current one. Once the group is set up, every time you exit Emacs, it copies the file you edited to the other files in its group. You can also do the copying without exiting Emacs, by typing M-x shadow-copy-files.

To set up a shadow file group, use M-x shadow-define-literal-group or M-x shadow-define-regexp-group. See their documentation strings for further information.

Before copying a file to its shadows, Emacs asks for confirmation. You can answer "no" to bypass copying of this file, this time. To cancel the shadowing permanently for a certain file, use M-x shadow-cancel to eliminate or change the shadow file group.

A shadow cluster is a group of hosts that share directories, so that copying to or from one of them is sufficient to update the file on all of them. Each shadow cluster has a name, and specifies the network address of a primary host (the one we copy files to), and a regular expression that matches the hostnames of all the other hosts in the cluster. You can define a shadow cluster with M-x shadow-define-cluster.