Mercurial > emacs
view etc/ONEWS.1 @ 44497:32d733dae726
(file-name-non-special): Handle return value of t from
`file-name-completion'.
| author | Andreas Schwab <schwab@suse.de> |
|---|---|
| date | Thu, 11 Apr 2002 12:13:53 +0000 |
| parents | a473fce1ed3a |
| children | 23a1cea22d13 |
line wrap: on
line source
Old GNU Emacs NEWS -- history of user-visible changes thru version 15. Copyright (C) 1985 Richard M. Stallman. See the end for copying conditions. Changes in Emacs 15 * Emacs now runs on Sun and Megatest 68000 systems; also on at least one 16000 system running 4.2. * Emacs now alters the output-start and output-stop characters to prevent C-s and C-q from being considered as flow control by cretinous rlogin software in 4.2. * It is now possible convert Mocklisp code (for Gosling Emacs) to Lisp code that can run in GNU Emacs. M-x convert-mocklisp-buffer converts the contents of the current buffer from Mocklisp to GNU Emacs Lisp. You should then save the converted buffer with C-x C-w under a name ending in ".el" There are probably some Mocklisp constructs that are not handled. If you encounter one, feel free to report the failure as a bug. The construct will be handled in a future Emacs release, if that is not not too hard to do. Note that lisp code converted from Mocklisp code will not necessarily run as fast as code specifically written for GNU Emacs, nor will it use the many features of GNU Emacs which are not present in Gosling's emacs. (In particular, the byte-compiler (m-x byte-compile-file) knows little about compilation of code directly converted from mocklisp.) It is envisaged that old mocklisp code will be incrementally converted to GNU lisp code, with M-x convert-mocklisp-buffer being the first step in this process. * Control-x n (narrow-to-region) is now by default a disabled command. This means that, if you issue this command, it will ask whether you really mean it. You have the opportunity to enable the command permanently at that time, so you will not be asked again. This will place the form "(put 'narrow-to-region 'disabled nil)" in your .emacs file. * Tags now prompts for the tag table file name to use. All the tags commands ask for the tag table file name if you have not yet specified one. Also, the command M-x visit-tag-table can now be used to specify the tag table file name initially, or to switch to a new tag table. * If truncate-partial-width-windows is non-nil (as it intially is), all windows less than the full screen width (that is, made by side-by-side splitting) truncate lines rather than continuing them. * Emacs now checks for Lisp stack overflow to avoid fatal errors. The depth in eval, apply and funcall may not exceed max-lisp-eval-depth. The depth in variable bindings and unwind-protects may not exceed max-specpdl-size. If either limit is exceeded, an error occurs. You can set the limits to larger values if you wish, but if you make them too large, you are vulnerable to a fatal error if you invoke Lisp code that does infinite recursion. * New hooks find-file-hook and write-file-hook. Both of these variables if non-nil should be functions of no arguments. At the time they are called (current-buffer) will be the buffer being read or written respectively. find-file-hook is called whenever a file is read into its own buffer, such as by calling find-file, revert-buffer, etc. It is not called by functions such as insert-file which do not read the file into a buffer of its own. find-file-hook is called after the file has been read in and its local variables (if any) have been processed. write-file-hook is called just before writing out a file from a buffer. * The initial value of shell-prompt-pattern is now "^[^#$%>]*[#$%>] *" * If the .emacs file sets inhibit-startup-message to non-nil, the messages normally printed by Emacs at startup time are inhibited. * Facility for run-time conditionalization on the basis of emacs features. The new variable features is a list of symbols which represent "features" of the executing emacs, for use in run-time conditionalization. The function featurep of one argument may be used to test for the presence of a feature. It is just the same as (not (null (memq FEATURE features))) where FEATURE is its argument. For example, (if (featurep 'magic-window-hack) (transmogrify-window 'vertical) (split-window-vertically)) The function provide of one argument "announces" that FEATURE is present. It is much the same as (if (not (featurep FEATURE)) (setq features (cons FEATURE features))) The function require with arguments FEATURE and FILE-NAME loads FILE-NAME (which should contain the form (provide FEATURE)) unless FEATURE is present. It is much the same as (if (not (featurep FEATURE)) (progn (load FILE-NAME) (if (not featurep FEATURE) (error ...)))) FILE-NAME is optional and defaults to FEATURE. * New function load-average. This returns a list of three integers, which are the current 1 minute, 5 minute and 15 minute load averages, each multiplied by a hundred (since normally they are floating point numbers). * Per-terminal libraries loaded automatically. Emacs when starting up on terminal type T automatically loads a library named term-T. T is the value of the TERM environment variable. Thus, on terminal type vt100, Emacs would do (load "term-vt100" t t). Such libraries are good places to set the character translation table. It is a bad idea to redefine lots of commands in a per-terminal library, since this affects all users. Instead, define a command to do the redefinitions and let the user's init file, which is loaded later, call that command or not, as the user prefers. * Programmer's note: detecting killed buffers. Buffers are eliminated by explicitly killing them, using the function kill-buffer. This does not eliminate or affect the pointers to the buffer which may exist in list structure. If you have a pointer to a buffer and wish to tell whether the buffer has been killed, use the function buffer-name. It returns nil on a killed buffer, and a string on a live buffer. * New ways to access the last command input character. The function last-key-struck, which used to return the last input character that was read by command input, is eliminated. Instead, you can find this information as the value of the variable last-command-char. (This variable used to be called last-key). Another new variable, last-input-char, holds the last character read from the command input stream regardless of what it was read for. last-input-char and last-command-char are different only inside a command that has called read-char to read input. * The new switch -kill causes Emacs to exit after processing the preceding command line arguments. Thus, emacs -l lib data -e do-it -kill means to load lib, find file data, call do-it on no arguments, and then exit. * The config.h file has been modularized. Options that depend on the machine you are running on are defined in a file whose name starts with "m-", such as m-vax.h. Options that depend on the operating system software version you are running on are defined in a file whose name starts with "s-", such as s-bsd4.2.h. config.h includes one m- file and one s- file. It also defines a few other options whose values do not follow from the machine type and system type being used. Installers normally will have to select the correct m- and s- files but will never have to change their contents. * Termcap AL and DL strings are understood. If the termcap entry defines AL and DL strings, for insertion and deletion of multiple lines in one blow, Emacs now uses them. This matters most on certain bit map display terminals for which scrolling is comparatively slow. * Bias against scrolling screen far on fast terminals. Emacs now prefers to redraw a few lines rather than shift them a long distance on the screen, when the terminal is fast. * New major mode, mim-mode. This major mode is for editing MDL code. Perhaps a MDL user can explain why it is not called mdl-mode. You must load the library mim-mode explicitly to use this. * GNU documentation formatter `texinfo'. The `texinfo' library defines a format for documentation files which can be passed through Tex to make a printed manual or passed through texinfo to make an Info file. Texinfo is documented fully by its own Info file; compare this file with its source, texinfo.texinfo, for additional guidance. All documentation files for GNU utilities should be written in texinfo input format. Tex processing of texinfo files requires the Botex macro package. This is not ready for distribution yet, but will appear at a later time. * New function read-from-string (emacs 15.29) read-from-string takes three arguments: a string to read from, and optionally start and end indices which delimit a substring from which to read. (They default to 0 and the length of the string, respectively.) This function returns a cons cell whose car is the object produced by reading from the string and whose cdr is a number giving the index in the string of the first character not read. That index may be passed as the second argument to a later call to read-from-string to read the next form represented by the string. In addition, the function read now accepts a string as its argument. In this case, it calls read-from-string on the whole string, and returns the car of the result. (ie the actual object read.) Changes in Emacs 14 * Completion now prints various messages such as [Sole Completion] or [Next Character Not Unique] to describe the results obtained. These messages appear after the text in the minibuffer, and remain on the screen until a few seconds go by or you type a key. * The buffer-read-only flag is implemented. Setting or binding this per-buffer variable to a non-nil value makes illegal any operation which would modify the textual content of the buffer. (Such operations signal a buffer-read-only error) The read-only state of a buffer may be altered using toggle-read-only (C-x C-q) The buffers used by Rmail, Dired, Rnews, and Info are now read-only by default to prevent accidental damage to the information in those buffers. * Functions car-safe and cdr-safe. These functions are like car and cdr when the argument is a cons. Given an argument not a cons, car-safe always returns nil, with no error; the same for cdr-safe. * The new function user-real-login-name returns the name corresponding to the real uid of the Emacs process. This is usually the same as what user-login-name returns; however, when Emacs is invoked from su, user-real-login-name returns "root" but user-login-name returns the name of the user who invoked su. Changes in Emacs 13 * There is a new version numbering scheme. What used to be the first version number, which was 1, has been discarded since it does not seem that I need three levels of version number. However, a new third version number has been added to represent changes by user sites. This number will always be zero in Emacs when I distribute it; it will be incremented each time Emacs is built at another site. * There is now a reader syntax for Meta characters: \M-CHAR means CHAR or'ed with the Meta bit. For example: ?\M-x is (+ ?x 128) ?\M-\n is (+ ?\n 128) ?\M-\^f is (+ ?\^f 128) This syntax can be used in strings too. Note, however, that Meta characters are not meaningful in key sequences being passed to define-key or lookup-key; you must use ESC characters (\e) in them instead. ?\C- can be used likewise for control characters. (13.9) * Installation change The string "../lisp" now adds to the front of the load-path used for searching for Lisp files during Emacs initialization. It used to replace the path specified in paths.h entirely. Now the directory ../lisp is searched first and the directoris specified in paths.h are searched afterward. Changes in Emacs 1.12 * There is a new installation procedure. See the file INSTALL that comes in the top level directory in the tar file or tape. * The Meta key is now supported on terminals that have it. This is a shift key which causes the high bit to be turned on in all input characters typed while it is held down. read-char now returns a value in the range 128-255 if a Meta character is typed. When interpreted as command input, a Meta character is equivalent to a two character sequence, the meta prefix character followed by the un-metized character (Meta-G unmetized is G). The meta prefix character is specified by the value of the variable meta-prefix-char. If this character (normally Escape) has been redefined locally with a non-prefix definition (such as happens in completing minibuffers) then the local redefinition is suppressed when the character is not the last one in a key sequence. So the local redefinition is effective if you type the character explicitly, but not effective if the character comes from the use of the Meta key. * `-' is no longer a completion command in the minibuffer. It is an ordinary self-inserting character. * The list load-path of directories load to search for Lisp files is now controlled by the EMACSLOADPATH environment variable [[ Note this was originally EMACS-LOAD-PATH and has been changed again; sh does not deal properly with hyphens in env variable names]] rather than the EPATH environment variable. This is to avoid conflicts with other Emacses. While Emacs is being built initially, the load-path is now just ("../lisp"), ignoring paths.h. It does not ignore EMACSLOADPATH, however; you should avoid having this variable set while building Emacs. * You can now specify a translation table for keyboard input characters, as a way of exchanging or substituting keys on the keyboard. If the value of keyboard-translate-table is a string, every character received from the keyboard is used as an index in that string, and the character at that index in the string is used as input instead of what was actually typed. If the actual input character is >= the length of the string, it is used unchanged. One way this feature can be used is to fix bad keyboard designes. For example, on some terminals, Delete is Shift-Underscore. Since Delete is a more useful character than Underscore, it is an improvement to make the unshifted character Delete and the shifted one Underscore. This can be done with ;; First make a translate table that does the identity translation. (setq keyboard-translate-table (make-string 128 0)) (let ((i 0)) (while (< i 128) (aset keyboard-translate-table i i) (setq i (1+ i)))) ;; Now alter translations of some characters. (aset keyboard-translate-table ?\_ ?\^?) (aset keyboard-translate-table ?\^? ?\_) If your terminal has a Meta key and can therefore send codes up to 255, Meta characters are translated through elements 128 through 255 of the translate table, and therefore are translated independently of the corresponding non-Meta characters. You must therefore establish translations independently for the Meta characters if you want them too: ;; First make a translate table that does the identity translation. (setq keyboard-translate-table (make-string 256 0)) (let ((i 0)) (while (< i 256) (aset keyboard-translate-table i i) (setq i (1+ i)))) ;; Now alter translations of some characters. (aset keyboard-translate-table ?\_ ?\^?) (aset keyboard-translate-table ?\^? ?\_) ;; Now alter translations of some Meta characters. (aset keyboard-translate-table (+ 128 ?\_) (+ 128 ?\^?)) (aset keyboard-translate-table (+ 128 ?\^?) (+ 128 ?\_)) * (process-kill-without-query PROCESS) This marks the process so that, when you kill Emacs, you will not on its account be queried about active subprocesses. Changes in Emacs 1.11 * The commands C-c and C-z have been interchanged, for greater compatibility with normal Unix usage. C-z now runs suspend-emacs and C-c runs exit-recursive-edit. * The value returned by file-name-directory now ends with a slash. (file-name-directory "foo/bar") => "foo/". This avoids confusing results when dealing with files in the root directory. The value of the per-buffer variable default-directory is also supposed to have a final slash now. * There are now variables to control the switches passed to `ls' by the C-x C-d command (list-directory). list-directory-brief-switches is a string, initially "-CF", used for brief listings, and list-directory-verbose-switches is a string, initially "-l", used for verbose ones. * For Ann Arbor Ambassador terminals, the termcap "ti" string is now used to initialize the screen geometry on entry to Emacs, and the "te" string is used to set it back on exit. If the termcap entry does not define the "ti" or "te" string, Emacs does what it used to do. Changes in Emacs 1.10 * GNU Emacs has been made almost 1/3 smaller. It now dumps out as only 530kbytes on Vax 4.2bsd. * The term "checkpoint" has been replaced by "auto save" throughout the function names, variable names and documentation of GNU Emacs. * The function load now tries appending ".elc" and ".el" to the specified filename BEFORE it tries the filename without change. * rmail now makes the mode line display the total number of messages and the current message number. The "f" command now means forward a message to another user. The command to search through all messages for a string is now "F". The "u" command now means to move back to the previous message and undelete it. To undelete the selected message, use Meta-u. * The hyphen character is now equivalent to a Space while in completing minibuffers. Both mean to complete an additional word. * The Lisp function error now takes args like format which are used to construct the error message. * Redisplay will refuse to start its display at the end of the buffer. It will pick a new place to display from, rather than use that. * The value returned by garbage-collect has been changed. Its first element is no longer a number but a cons, whose car is the number of cons cells now in use, and whose cdr is the number of cons cells that have been made but are now free. The second element is similar but describes symbols rather than cons cells. The third element is similar but describes markers. * The variable buffer-name has been eliminated. The function buffer-name still exists. This is to prevent user programs from changing buffer names without going through the rename-buffer function. Changes in Emacs 1.9 * When a fill prefix is in effect, paragraphs are started or separated by lines that do not start with the fill prefix. Also, a line which consists of the fill prefix followed by white space separates paragraphs. * C-x C-v runs the new function find-alternate-file. It finds the specified file, switches to that buffer, and kills the previous current buffer. (It requires confirmation if that buffer had changes.) This is most useful after you find the wrong file due to a typo. * Exiting the minibuffer moves the cursor to column 0, to show you that it has really been exited. * Meta-g (fill-region) now fills each paragraph in the region individually. To fill the region as if it were a single paragraph (for when the paragraph-delimiting mechanism does the wrong thing), use fill-region-as-paragraph. * Tab in text mode now runs the function tab-to-tab-stop. A new mode called indented-text-mode is like text-mode except that in it Tab runs the function indent-relative, which indents the line under the previous line. If auto fill is enabled while in indented-text-mode, the new lines that it makes are indented. * Functions kill-rectangle and yank-rectangle. kill-rectangle deletes the rectangle specified by dot and mark (or by two arguments) and saves it in the variable killed-rectangle. yank-rectangle inserts the rectangle in that variable. Tab characters in a rectangle being saved are replaced by spaces in such a way that their appearance will not be changed if the rectangle is later reinserted at a different column position. * `+' in a regular expression now means to repeat the previous expression one or more times. `?' means to repeat it zero or one time. They are in all regards like `*' except for the number of repetitions they match. \< in a regular expression now matches the null string when it is at the beginning of a word; \> matches the null string at the end of a word. * C-x p narrows the buffer so that only the current page is visible. * C-x ) with argument repeats the kbd macro just defined that many times, counting the definition as one repetition. * C-x ( with argument begins defining a kbd macro starting with the last one defined. It executes that previous kbd macro initially, just as if you began by typing it over again. * C-x q command queries the user during kbd macro execution. With prefix argument, enters recursive edit, reading keyboard commands even within a kbd macro. You can give different commands each time the macro executes. Without prefix argument, reads a character. Your options are: Space -- execute the rest of the macro. Delete -- skip the rest of the macro; start next repetition. C-d -- skip rest of the macro and don't repeat it any more. C-r -- enter a recursive edit, then on exit ask again for a character C-l -- redisplay screen and ask again." * write-kbd-macro and append-kbd-macro are used to save a kbd macro definition in a file (as Lisp code to redefine the macro when the file is loaded). These commands differ in that write-kbd-macro discards the previous contents of the file. If given a prefix argument, both commands record the keys which invoke the macro as well as the macro's definition. * The variable global-minor-modes is used to display strings in the mode line of all buffers. It should be a list of elements thaht are conses whose cdrs are strings to be displayed. This complements the variable minor-modes, which has the same effect but has a separate value in each buffer. * C-x = describes horizontal scrolling in effect, if any. * Return now auto-fills the line it is ending, in auto fill mode. Space with zero as argument auto-fills the line before it just like Space without an argument. Changes in Emacs 1.8 This release mostly fixes bugs. There are a few new features: * apropos now sorts the symbols before displaying them. Also, it returns a list of the symbols found. apropos now accepts a second arg PRED which should be a function of one argument; if PRED is non-nil, each symbol is tested with PRED and only symbols for which PRED returns non-nil appear in the output or the returned list. If the third argument to apropos is non-nil, apropos does not display anything; it merely returns the list of symbols found. C-h a now runs the new function command-apropos rather than apropos, and shows only symbols with definitions as commands. * M-x shell sends the command if (-f ~/.emacs_NAME)source ~/.emacs_NAME invisibly to the shell when it starts. Here NAME is replaced by the name of shell used, as it came from your ESHELL or SHELL environment variable but with directory name, if any, removed. * M-, now runs the command tags-loop-continue, which is used to resume a terminated tags-search or tags-query-replace. Changes in Emacs 1.7 It's Beat CCA Week. * The initial buffer is now called "*scratch*" instead of "scratch", so that all buffer names used automatically by Emacs now have *'s. * Undo information is now stored separately for each buffer. The Undo command (C-x u) always applies to the current buffer only. C-_ is now a synonym for C-x u. (buffer-flush-undo BUFFER) causes undo information not to be kept for BUFFER, and frees the space that would have been used to hold it. In any case, no undo information is kept for buffers whose names start with spaces. (These buffers also do not appear in the C-x C-b display.) * Rectangle operations are now implemented. C-x r stores the rectangle described by dot and mark into a register; it reads the register name from the keyboard. C-x g, the command to insert the contents of a register, can be used to reinsert the rectangle elsewhere. Other rectangle commands include open-rectangle: insert a blank rectangle in the position and size described by dot and mark, at its corners; the existing text is pushed to the right. clear-rectangle: replace the rectangle described by dot ane mark with blanks. The previous text is deleted. delete-rectangle: delete the text of the specified rectangle, moving the text beyond it on each line leftward. * Side-by-side windows are allowed. Use C-x 5 to split the current window into two windows side by side. C-x } makes the selected window ARG columns wider at the expense of the windows at its sides. C-x { makes the selected window ARG columns narrower. An argument to C-x 5 specifies how many columns to give to the leftmost of the two windows made. C-x 2 now accepts a numeric argument to specify the number of lines to give to the uppermost of the two windows it makes. * Horizontal scrolling of the lines in a window is now implemented. C-x < (scroll-left) scrolls all displayed lines left, with the numeric argument (default 1) saying how far to scroll. When the window is scrolled left, some amount of the beginning of each nonempty line is replaced by an "$". C-x > scrolls right. If a window has no text hidden at the left margin, it cannot be scrolled any farther right than that. When nonzero leftwards scrolling is in effect in a window. lines are automatically truncated at the window's right margin regardless of the value of the variable truncate-lines in the buffer being displayed. * C-x C-d now uses the default output format of `ls', which gives just file names in multiple columns. C-u C-x C-d passes the -l switch to `ls'. * C-t at the end of a line now exchanges the two preceding characters. All the transpose commands now interpret zero as an argument to mean to transpose the textual unit after or around dot with the one after or around the mark. * M-! executes a shell command in an inferior shell and displays the output from it. With a prefix argument, it inserts the output in the current buffer after dot and sets the mark after the output. The shell command gets /dev/null as its standard input. M-| is like M-! but passes the contents of the region as input to the shell command. A prefix argument makes the output from the command replace the contents of the region. * The mode line will now say "Def" after the major mode while a keyboard macro is being defined. * The variable fill-prefix is now used by Meta-q. Meta-q removes the fill prefix from lines that start with it before filling, and inserts the fill prefix on each line after filling. The command C-x . sets the fill prefix equal to the text on the current line before dot. * The new command Meta-j (indent-new-comment-line), is like Linefeed (indent-new-line) except when dot is inside a comment; in that case, Meta-j inserts a comment starter on the new line, indented under the comment starter above. It also inserts a comment terminator at the end of the line above, if the language being edited calls for one. * Rmail should work correctly now, and has some C-h m documentation. Changes in Emacs 1.6 * save-buffers-kill-emacs is now on C-x C-c while C-x C-z does suspend-emacs. This is to make C-x C-c like the normal Unix meaning of C-c and C-x C-z linke the normal Unix meaning of C-z. * M-ESC (eval-expression) is now a disabled command by default. This prevents users who type ESC ESC accidentally from getting confusing results. Put (put 'eval-expression 'disabled nil) in your ~/.emacs file to enable the command. * Self-inserting text is grouped into bunches for undoing. Each C-x u command undoes up to 20 consecutive self-inserting characters. * Help f now uses as a default the function being called in the innermost Lisp expression that dot is in. This makes it more convenient to use while writing Lisp code to run in Emacs. (If the text around dot does not appear to be a call to a Lisp function, there is no default.) Likewise, Help v uses the symbol around or before dot as a default, if that is a variable name. * Commands that read filenames now insert the default directory in the minibuffer, to become part of your input. This allows you to see what the default is. You may type a filename which goes at the end of the default directory, or you may edit the default directory as you like to create the input you want to give. You may also type an absolute pathname (starting with /) or refer to a home directory (input starting with ~) after the default; the presence of // or /~ causes everything up through the slash that precedes your type-in to be ignored. Returning the default directory without change, including the terminating slash, requests the use of the default file name (usually the visited file's name). Set the variable insert-default-directory to nil to turn off this feature. * M-x shell now uses the environment variable ESHELL, if it exists, as the file name of the shell to run. If there is no ESHELL variable, the SHELL variable is used. This is because some shells do not work properly as inferiors of Emacs (or anything like Emacs). * A new variable minor-modes now exists, with a separate value in each buffer. Its value should be an alist of elements (MODE-FUNCTION-SYMBOL . PRETTY-NAME-STRING), one for each minor mode that is turned on in the buffer. The pretty name strings are displayed in the mode line after the name of the major mode (with spaces between them). The mode function symbols should be symbols whose function definitions will turn on the minor mode if given 1 as an argument; they are present so that Help m can find their documentation strings. * The format of tag table files has been changed. The new format enables Emacs to find tags much faster. A new program, etags, exists to make the kind of tag table that Emacs wants. etags is invoked just like ctags; in fact, if you give it any switches, it does exactly what ctags would do. Give it the empty switch ("-") to make it act like ctags with no switches. etags names the tag table file "TAGS" rather than "tags", so that these tag tables and the standard Unix ones can coexist. The tags library can no longer use standard ctags-style tag tables files. * The file of Lisp code Emacs reads on startup is now called ~/.emacs rather than ~/.emacs_pro. * copy-file now gives the copied file the same mode bits as the original file. * Output from a process inserted into the process's buffer no longer sets the buffer's mark. Instead it sets a marker associated with the process to point to the end of the inserted text. You can access this marker with (process-mark PROCESS) and then either examine its position with marker-position or set its position with set-marker. * completing-read takes a new optional fifth argument which, if non-nil, should be a string of text to insert into the minibuffer before reading user commands. * The Lisp function elt now exists: (elt ARRAY N) is like (aref ARRAY N), (elt LIST N) is like (nth N LIST). * rplaca is now a synonym for setcar, and rplacd for setcdr. eql is now a synonym for eq; it turns out that the Common Lisp distinction between eq and eql is insignificant in Emacs. numberp is a new synonym for integerp. * auto-save has been renamed to auto-save-mode. * Auto save file names for buffers are now created by the function make-auto-save-file-name. This is so you can redefine that function to change the way auto save file names are chosen. * expand-file-name no longer discards a final slash. (expand-file-name "foo" "/lose") => "/lose/foo" (expand-file-name "foo/" "/lose") => "/lose/foo/" Also, expand-file-name no longer substitutes $ constructs. A new function substitute-in-file-name does this. Reading a file name with read-file-name or the `f' or`F' option of interactive calling uses substitute-in-file-name on the file name that was read and returns the result. All I/O primitives including insert-file-contents and delete-file call expand-file-name on the file name supplied. This change makes them considerably faster in the usual case. * Interactive calling spec strings allow the new code letter 'D' which means to read a directory name. It is like 'f' except that the default if the user makes no change in the minibuffer is to return the current default directory rather than the current visited file name. Changes in Emacs 1.5 * suspend-emacs now accepts an optional argument which is a string to be stuffed as terminal input to be read by Emacs's superior shell after Emacs exits. A library called ledit exists which uses this feature to transmit text to a Lisp job running as a sibling of Emacs. * If find-file is given the name of a directory, it automatically invokes dired on that directory rather than reading in the binary data that make up the actual contents of the directory according to Unix. * Saving an Emacs buffer now preserves the file modes of any previously existing file with the same name. This works using new Lisp functions file-modes and set-file-modes, which can be used to read or set the mode bits of any file. * The Lisp function cond now exists, with its traditional meaning. * defvar and defconst now permit the documentation string to be omitted. defvar also permits the initial value to be omitted; then it acts only as a comment. Changes in Emacs 1.4 * Auto-filling now normally indents the new line it creates by calling indent-according-to-mode. This function, meanwhile, has in Fundamental and Text modes the effect of making the line have an indentation of the value of left-margin, a per-buffer variable. Tab no longer precisely does indent-according-to-mode; it does that in all modes that supply their own indentation routine, but in Fundamental, Text and allied modes it inserts a tab character. * The command M-x grep now invokes grep (on arguments supplied by the user) and reads the output from grep asynchronously into a buffer. The command C-x ` can be used to move to the lines that grep has found. This is an adaptation of the mechanism used for running compilations and finding the loci of error messages. You can now use C-x ` even while grep or compilation is proceeding; as more matches or error messages arrive, C-x ` will parse them and be able to find them. * M-x mail now provides a command to send the message and "exit"--that is, return to the previously selected buffer. It is C-z C-z. * Tab in C mode now tries harder to adapt to all indentation styles. If the line being indented is a statement that is not the first one in the containing compound-statement, it is aligned under the beginning of the first statement. * The functions screen-width and screen-height return the total width and height of the screen as it is now being used. set-screen-width and set-screen-height tell Emacs how big to assume the screen is; they each take one argument, an integer. * The Lisp function 'function' now exists. function is the same as quote, except that it serves as a signal to the Lisp compiler that the argument should be compiled as a function. Example: (mapcar (function (lambda (x) (+ x 5))) list) * The function set-key has been renamed to global-set-key. undefine-key and local-undefine-key has been renamed to global-unset-key and local-unset-key. * Emacs now collects input from asynchronous subprocesses while waiting in the functions sleep-for and sit-for. * Shell mode's Newline command attempts to distinguish subshell prompts from user input when issued in the middle of the buffer. It no longer reexecutes from dot to the end of the line; it reeexecutes the entire line minus any prompt. The prompt is recognized by searching for the value of shell-prompt-pattern, starting from the beginning of the line. Anything thus skipped is not reexecuted. Changes in Emacs 1.3 * An undo facility exists now. Type C-x u to undo a batch of changes (usually one command's changes, but some commands such as query-replace divide their changes into multiple batches. You can repeat C-x u to undo further. As long as no commands other than C-x u intervene, each one undoes another batch. A numeric argument to C-x u acts as a repeat count. If you keep on undoing, eventually you may be told that you have used up all the recorded undo information. Some actions, such as reading in files, discard all undo information. The undo information is not currently stored separately for each buffer, so it is mainly good if you do something totally spastic. [This has since been fixed.] * A learn-by-doing tutorial introduction to Emacs now exists. Type C-h t to enter it. * An Info documentation browser exists. Do M-x info to enter it. It contains a tutorial introduction so that no more documentation is needed here. As of now, the only documentation in it is that of Info itself. * Help k and Help c are now different. Help c prints just the name of the function which the specified key invokes. Help k prints the documentation of the function as well. * A document of the differences between GNU Emacs and Twenex Emacs now exists. It is called DIFF, in the same directory as this file. * C mode can now indent comments better, including multi-line ones. Meta-Control-q now reindents comment lines within the expression being aligned. * Insertion of a close-parenthesis now shows the matching open-parenthesis even if it is off screen, by printing the text following it on its line in the minibuffer. * A file can now contain a list of local variable values to be in effect when the file is edited. See the file DIFF in the same directory as this file for full details. * A function nth is defined. It means the same thing as in Common Lisp. * The function install-command has been renamed to set-key. It now takes the key sequence as the first argument and the definition for it as the second argument. Likewise, local-install-command has been renamed to local-set-key. Changes in Emacs 1.2 * A Lisp single-stepping and debugging facility exists. To cause the debugger to be entered when an error occurs, set the variable debug-on-error non-nil. To cause the debugger to be entered whenever function foo is called, do (debug-on-entry 'foo). To cancel this, do (cancel-debug-on-entry 'foo). debug-on-entry does not work for primitives (written in C), only functions written in Lisp. Most standard Emacs commands are in Lisp. When the debugger is entered, the selected window shows a buffer called " *Backtrace" which displays a series of stack frames, most recently entered first. For each frame, the function name called is shown, usually followed by the argument values unless arguments are still being calculated. At the beginning of the buffer is a description of why the debugger was entered: function entry, function exit, error, or simply that the user called the function `debug'. To exit the debugger and return to top level, type `q'. In the debugger, you can evaluate Lisp expressions by typing `e'. This is equivalent to `M-ESC'. When the debugger is entered due to an error, that is all you can do. When it is entered due to function entry (such as, requested by debug-on-entry), you have two options: Continue execution and reenter debugger after the completion of the function being entered. Type `c'. Continue execution but enter the debugger before the next subexpression. Type `d'. You will see that some stack frames are marked with *. This means the debugger will be entered when those frames exit. You will see the value being returned in the first line of the backtrace buffer. Your options: Continue execution, and return that value. Type `c'. Continue execution, and return a specified value. Type `r'. You can mark a frame to enter the debugger on exit with the `b' command, or clear such a mark with `u'. * Lisp macros now exist. For example, you can write (defmacro cadr (arg) (list 'car (list 'cdr arg))) and then the expression (cadr foo) will expand into (car (cdr foo)) Changes in Emacs 1.1 * The initial buffer is now called "scratch" and is in a new major mode, Lisp Interaction mode. This mode is intended for typing Lisp expressions, evaluating them, and having the values printed into the buffer. Type Linefeed after a Lisp expression, to evaluate the expression and have its value printed into the buffer, advancing dot. The other commands of Lisp mode are available. * The C-x C-e command for evaluating the Lisp expression before dot has been changed to print the value in the minibuffer line rather than insert it in the buffer. A numeric argument causes the printed value to appear in the buffer instead. * In Lisp mode, the command M-C-x evaluates the defun containing or following dot. The value is printed in the minibuffer. * The value of a Lisp expression evaluated using M-ESC is now printed in the minibuffer. * M-q now runs fill-paragraph, independent of major mode. * C-h m now prints documentation on the current buffer's major mode. What it prints is the documentation of the major mode name as a function. All major modes have been equipped with documentation that describes all commands peculiar to the major mode, for this purpose. * You can display a Unix manual entry with the M-x manual-entry command. * You can run a shell, displaying its output in a buffer, with the M-x shell command. The Return key sends input to the subshell. Output is printed inserted automatically in the buffer. Commands C-c, C-d, C-u, C-w and C-z are redefined for controlling the subshell and its subjobs. "cd", "pushd" and "popd" commands are recognized as you enter them, so that the default directory of the Emacs buffer always remains the same as that of the subshell. * C-x $ (that's a real dollar sign) controls line-hiding based on indentation. With a numeric arg N > 0, it causes all lines indented by N or more columns to become invisible. They are, effectively, tacked onto the preceding line, where they are represented by " ..." on the screen. (The end of the preceding visible line corresponds to a screen cursor position before the "...". Anywhere in the invisible lines that follow appears on the screen as a cursor position after the "...".) Currently, all editing commands treat invisible lines just like visible ones, except for C-n and C-p, which have special code to count visible lines only. C-x $ with no argument turns off this mode, which in any case is remembered separately for each buffer. * Outline mode is another form of selective display. It is a major mode invoked with M-x outline-mode. It is intended for editing files that are structured as outlines, with heading lines (lines that begin with one or more asterisks) and text lines (all other lines). The number of asterisks in a heading line are its level; the subheadings of a heading line are all following heading lines at higher levels, until but not including the next heading line at the same or a lower level, regardless of intervening text lines. In outline mode, you have commands to hide (remove from display) or show the text or subheadings under each heading line independently. Hidden text or subheadings are invisibly attached to the end of the preceding heading line, so that if you kill the hading line and yank it back elsewhere all the invisible lines accompany it. All editing commands treat hidden outline-mode lines as part of the preceding visible line. * C-x C-z runs save-buffers-kill-emacs offers to save each file buffer, then exits. * C-c's function is now called suspend-emacs. * The command C-x m runs mail, which switches to a buffer *mail* and lets you compose a message to send. C-x 4 m runs mail in another window. Type C-z C-s in the mail buffer to send the message according to what you have entered in the buffer. You must separate the headers from the message text with an empty line. * You can now dired partial directories (specified with names containing *'s, etc, all processed by the shell). Also, you can dired more than one directory; dired names the buffer according to the filespec or directory name. Reinvoking dired on a directory already direded just switches back to the same directory used last time; do M-x revert if you want to read in the current contents of the directory. C-x d runs dired, and C-x 4 d runs dired in another window. C-x C-d (list-directory) also allows partial directories now. Lisp programming changes * t as an output stream now means "print to the minibuffer". If there is already text in the minibuffer printed via t as an output stream, the new text is appended to the old (or is truncated and lost at the margin). If the minibuffer contains text put there for some other reason, it is cleared first. t is now the top-level value of standard-output. t as an input stream now means "read via the minibuffer". The minibuffer is used to read a line of input, with editing, and this line is then parsed. Any excess not used by `read' is ignored; each `read' from t reads fresh input. t is now the top-level value of standard-input. * A marker may be used as an input stream or an output stream. The effect is to grab input from where the marker points, advancing it over the characters read, or to insert output at the marker and advance it. * Output from an asynchronous subprocess is now inserted at the end of the associated buffer, not at the buffer's dot, and the buffer's mark is set to the end of the inserted output each time output is inserted. * (pos-visible-in-window-p POS WINDOW) returns t if position POS in WINDOW's buffer is in the range that is being displayed in WINDOW; nil if it is scrolled vertically out of visibility. If display in WINDOW is not currently up to date, this function calculates carefully whether POS would appear if display were done immediately based on the current (window-start WINDOW). POS defaults to (dot), and WINDOW to (selected-window). * Variable buffer-alist replaced by function (buffer-list). The actual alist of buffers used internally by Emacs is now no longer accessible, to prevent the user from crashing Emacs by modifying it. The function buffer-list returns a list of all existing buffers. Modifying this list cannot hurt anything as a new list is constructed by each call to buffer-list. * load now takes an optional third argument NOMSG which, if non-nil, prevents load from printing a message when it starts and when it is done. * byte-recompile-directory is a new function which finds all the .elc files in a directory, and regenerates each one which is older than the corresponding .el (Lisp source) file. ---------------------------------------------------------------------- Copyright information: Copyright (C) 1985 Richard M. Stallman Permission is granted to anyone to make or distribute verbatim copies of this document as received, in any medium, provided that the copyright notice and this permission notice are preserved, thus giving the recipient permission to redistribute in turn. Permission is granted to distribute modified versions of this document, or of portions of it, under the above conditions, provided also that they carry prominent notices stating who last changed them. Local variables: mode: text end: