emacs/etc/NEWS.1-17

GNU Emacs NEWS -- history of user-visible changes.  26-Mar-1986

Copyright (C) 1985-1986, 2006-2023 Free Software Foundation, Inc.
See the end of the file for license conditions.


This file is about changes in Emacs versions 1 through 17.



* Changes in Emacs 17

** Frustrated?

Try M-x doctor.

** Bored?

Try M-x hanoi.

** Brain-damaged?

Try M-x yow.

** Sun3, Tahoe, Apollo, HP9000s300, Celerity, NCR Tower 32,
  Sequent, Stride, Encore, Plexus and AT&T 7300 machines supported.

The Tahoe, Sun3, Sequent and Celerity use 4.2.  In regard to the
Apollo, see the file APOLLO in this directory.  NCR Tower32,
HP9000s300, Stride and Nu run forms of System V.  System V rel 2 also
works on Vaxes now.  See etc/MACHINES.

** System V Unix supported, including subprocesses.

It should be possible now to bring up Emacs on a machine running
mere unameliorated system V Unix with no major work; just possible bug
fixes.  But you can expect to find a handful of those on any machine
that Emacs has not been run on before.

** Berkeley 4.1 Unix supported.

See etc/MACHINES.

** Portable 'alloca' provided.

Emacs can now run on machines that do not and cannot support the library
subroutine 'alloca' in the canonical fashion, using an 'alloca' emulation
written in C.

** On-line manual.

Info now contains an Emacs manual, with essentially the same text
as in the printed manual.

The manual can now be printed with a standard TeX.

Nicely typeset and printed copies of the manual are available
from the Free Software Foundation.

** Backup file version numbers.

Emacs now supports version numbers in backup files.

The first time you save a particular file in one editing session,
the old file is copied or renamed to serve as a backup file.
In the past, the name for the backup file was made by appending `~'
to the end of the original file name.

Now the backup file name can instead be made by appending ".~NN~" to
the original file name, where NN stands for a numeric version.  Each
time this is done, the new version number is one higher than the
highest previously used.

Thus, the active, current file does not have a version number.
Only the backups have them.

This feature is controlled by the variable 'version-control'.  If it
is 'nil', as normally, then numbered backups are made only for files
that already have numbered backups.  Backup names with just `~' are
used for files that have no numbered backups.

If 'version-control' is 'never', then the backup file's name is
made with just `~' in any case.

If 'version-control' is not 'nil' or 'never', numbered backups are
made unconditionally.

To prevent unlimited consumption of disk space, Emacs can delete
old backup versions automatically.  Generally Emacs keeps the first
few backups and the latest few backups, deleting any in between.
This happens every time a new backup is made.  The two variables that
control the deletion are 'kept-old-versions' and 'kept-new-versions'.
Their values are, respectively, the number of oldest backups to keep
and the number of newest ones to keep, each time a new backup is made.
The value of 'kept-new-versions' includes the backup just created.
By default, both values are 2.

If 'trim-versions-without-asking' is non-'nil', the excess middle versions
are deleted without a murmur.  If it is 'nil', the default, then you
are asked whether the excess middle versions should really be deleted.

Dired has a new command `.' which marks for deletion all but the latest
and oldest few of every numeric series of backups.  'kept-old-versions'
controls the number of oldest versions to keep, and 'dired-kept-versions'
controls the number of latest versions to keep.  A numeric argument to
the `.' command, if positive, specifies the number of latest versions
to keep, overriding 'dired-kept-versions'.  A negative argument specifies
the number of oldest versions to keep, using minus the argument to override
'kept-old-versions'.

** Immediate conflict detection.

Emacs now locks the files it is modifying, so that if
you start to modify within Emacs a file that is being
modified in another Emacs, you get an immediate warning.

The warning gives you three choices:
1. Give up, and do not make any changes.
2. Make changes anyway at your own risk.
3. Make changes anyway, and record yourself as
 the person locking the file (instead of whoever
 was previously recorded.)

Just visiting a file does not lock it.  It is locked
when you try to change the buffer that is visiting the file.
Saving the file unlocks it until you make another change.

Locking is done by writing a lock file in a special designated
directory.  If such a directory is not provided and told to
Emacs as part of configuring it for your machine, the lock feature
is turned off.

** M-x recover-file.

This command is used to get a file back from an auto-save
(after a system crash, for example).  It takes a file name
as argument and visits that file, but gets the data from the
file's last auto save rather than from the file itself.

** M-x normal-mode.

This command resets the current buffer's major mode and local
variables to be as specified by the visit filename, the -*- line
and/or the Local Variables: block at the end of the buffer.
It is the same thing normally done when a file is first visited.

** Echo area messages disappear shortly if minibuffer is in use.

Any message in the echo area disappears after 2 seconds
if the minibuffer is active.  This allows the minibuffer
to become visible again.

** C-z on System V runs a subshell.

On systems which do not allow programs to be suspended, the C-z command
forks a subshell that talks directly to the terminal, and then waits
for the subshell to exit.  This gets almost the effect of suspending
in that you can run other programs and then return to Emacs.  However,
you cannot log out from the subshell.

** C-c is always a prefix character.

Also, subcommands of C-c which are letters are always
reserved for the user.  No standard Emacs major mode
defines any of them.

** Picture mode C-c commands changed.

The old C-c k command is now C-c C-w.
The old C-c y command is now C-c C-x.

** Shell mode commands changed.

All the special commands of Shell mode are now moved onto
the C-c prefix.  Most are not changed aside from that.
Thus, the old Shell mode C-c command (kill current job)
is now C-c C-c; the old C-z (suspend current job) is now C-c C-z,
etc.

The old C-x commands are now C-c commands.  C-x C-k (kill output)
is now C-c C-o, and C-x C-v (show output) is now C-c C-r.

The old M-= (copy previous input) command is now C-c C-y.

** Shell mode recognizes aliases for 'pushd', 'popd' and 'cd'.

Shell mode now uses the variable 'shell-pushd-regexp' as a
regular expression to recognize any command name that is
equivalent to a 'pushd' command.  By default it is set up
to recognize just 'pushd' itself.  If you use aliases for
'pushd', change the regexp to recognize them as well.

There are also 'shell-popd-regexp' to recognize commands
with the effect of a 'popd', and 'shell-cd-regexp' to recognize
commands with the effect of a 'cd'.

** "Exit" command in certain modes now C-c C-c.

These include electric buffer menu mode, electric command history
mode, Info node edit mode, and Rmail edit mode.  In all these
modes, the command to exit used to be just C-c.

** Outline mode changes.

Lines that are not heading lines are now called "body" lines.
The command 'hide-text' is renamed to 'hide-body'.
The key M-H is renamed to C-c C-h.
The key M-S is renamed to C-c C-s.
The key M-s is renamed to C-c C-i.

Changes of line visibility are no longer undoable.  As a result,
they no longer use up undo memory and no longer interfere with
undoing earlier commands.

** Rmail changes.

The s and q commands now both expunge deleted messages before saving;
use C-x C-s to save without expunging.

The u command now undeletes the current message if it is deleted;
otherwise, it backs up as far as necessary to reach a deleted message,
and undeletes that one.  The u command in the summary behaves likewise,
but considers only messages listed in the summary.  The M-u command
has been eliminated.

The o and C-o keys' meanings are interchanged.
o now outputs to an Rmail file, and C-o to a Unix mail file.

The F command (rmail-find) is renamed to M-s (rmail-search).
Various new commands and features exist; see the Emacs manual.

** Local bindings described first in 'describe-bindings'.

** [...], {...} now balance in Fundamental mode.

** Nroff mode and TeX mode.

There are two new major modes for editing nroff input and TeX input.
See the Emacs manual for full information.

** New C indentation style variable 'c-brace-imaginary-offset'.

The value of 'c-brace-imaginary-offset', normally zero, controls the
indentation of a statement inside a brace-group where the open-brace
is not the first thing on a line.  The value says where the open-brace
is imagined to be, relative to the first nonblank character on the line.

** Dired improvements.

Dired now normally keeps the cursor at the beginning of the file name,
not at the beginning of the line.  The most used motion commands are
redefined in Dired to position the cursor this way.

'n' and 'p' are now equivalent in dired to 'C-n' and 'C-p'.

If any files to be deleted cannot be deleted, their names are
printed in an error message.

If the 'v' command is invoked on a file which is a directory,
dired is run on that directory.

** 'visit-tag-table' renamed 'visit-tags-table'.

This is so apropos of 'tags' finds everything you need to
know about in connection with Tags.

** 'mh-e' library uses C-c as prefix.

All the special commands of 'mh-rmail' now are placed on a
C-c prefix rather than on the C-x prefix.  This is for
consistency with other special modes with their own commands.

** M-$ or 'spell-word' checks word before point.

It used to check the word after point.

** Quitting during autoloading no longer causes trouble.

Now, when a file is autoloaded, all function redefinitions
and 'provide' calls are recorded and are undone if you quit
before the file is finished loading.

As a result, it no longer happens that some of the entry points
which are normally autoloading have been defined already, but the
entire file is not really present to support them.

** 'else' can now be indented correctly in C mode.

TAB in C mode now knows which 'if' statement an 'else' matches
up with, and can indent the 'else' correctly under the 'if',
even if the 'if' contained such things as another 'if' statement,
or a 'while' or 'for' statement, with no braces around it.

** 'batch-byte-compile'

Runs byte-compile-file on the files specified on the command line.
All the rest of the command line arguments are taken as files to
compile (or, if directories, to do byte-recompile-directory on).
Must be used only with -batch, and kills emacs on completion.
Each file will be processed even if an error occurred previously.
For example, invoke `emacs -batch -f batch-byte-compile *.el'.

** '-batch' changes.

'-batch' now implies '-q': no init file is loaded by Emacs when
'-batch' is used.  Also, no `term/TERMTYPE.el' file is loaded.  Auto
saving is not done except in buffers in which it is explicitly
requested.  Also, many echo-area printouts describing what is going on
are inhibited in batch mode, so that the only output you get is the
output you program specifically.

One echo-area message that is not suppressed is the one that says
that a file is being loaded.  That is because you can prevent this
message by passing 't' as the third argument to 'load'.

** Display of search string in incremental search.

Now, when you type C-s or C-r to reuse the previous search
string, that search string is displayed immediately in the echo area.

Three dots are displayed after the search string while search
is actually going on.

** View commands.

The commands C-x ], C-x [, C-x /, C-x j and C-x o are now
available inside 'view-buffer' and 'view-file', with their
normal meanings.

** Full-width windows preferred.

The 'other-window' commands prefer other full width windows,
and will split only full width windows.

** M-x rename-file can copy if necessary.

When used between different file systems, since actual renaming does
not work, the old file will be copied and deleted.

** Within C-x ESC, you can pick the command to repeat.

While editing a previous command to be repeated, inside C-x ESC,
you can now use the commands M-p and M-n to pick an earlier or
later command to repeat.  M-n picks the next earlier command
and M-p picks the next later one.  The new command appears in
the minibuffer, and you can go ahead and edit it, and repeat it
when you exit the minibuffer.

Using M-n or M-p within C-x ESC is like having used a different
numeric argument when you ran C-x ESC in the first place.

The command you finally execute using C-x ESC is added to the
front of the command history, unless it is identical with the
first thing in the command history.

** Use C-c C-c to exit from editing within Info.

It used to be C-z for this.  Somehow this use of C-z was
left out when all the others were moved.  The intention is that
C-z should always suspend Emacs.

** Default arg to C-x < and C-x > now window width minus 2.

These commands, which scroll the current window horizontally
by a specified number of columns, now scroll a considerable
distance rather than a single column if used with no argument.

** Auto Save Files Deleted.

The default value of 'delete-auto-save-files' is now 't', so that
when you save a file for real, its auto save file is deleted.

** Rnews changes.

The N, P and J keys in Rnews are renamed to M-n, M-p and M-j.
These keys move among newsgroups.

The n and p keys for moving sequentially between news articles now
accept repeat count arguments, and the + and - keys, made redundant by
this change, are eliminated.

The s command for outputting the current article to a file
is renamed as o, to be compatible with Rmail.

** Sendmail changes.

If you have a ~/.mailrc file, Emacs searches it for mailing address
aliases, and these aliases are expanded when you send mail in Emacs.

Fcc fields can now be used in the headers in the *mail* buffer
to specify files in which copies of the message should be put.
The message is written into those files in Unix mail file format.
The message as sent does not contain any Fcc fields in its header.
You can use any number of Fcc fields, but only one file name in each one.
The variable 'mail-archive-file-name', if non-'nil', can be a string
which is a file name; an Fcc to that file will be inserted in every
message when you begin to compose it.

A new command C-c q now exists in Mail mode.  It fills the
paragraphs of an old message that had been inserted with C-c y.

When the *mail* buffer is put in Mail mode, 'text-mode-hook'
is now run in addition to 'mail-mode-hook'.  text-mode-hook
is run first.

The new variable 'mail-header-separator' now specifies the string
to use on the line that goes between the headers and the message text.
By default it is still "--text follows this line--".

** Command history truncated automatically.

Just before each garbage collection, all but the last 30 elements
of the command history are discarded.


* Incompatible Lisp Programming Changes in Emacs 17

** `&quote' no longer supported.

This feature, which allowed Lisp functions to take arguments
that were not evaluated, has been eliminated, because it is
inescapably hard to make the compiler work properly with such
functions.

You should use macros instead.  A simple way to change any
code that uses `&quote' is to replace

   (defun foo (&quote x y z) ...

with

   (defmacro foo (x y z)
     (list 'foo-1 (list 'quote x) (list 'quote y) (list 'quote z)))

   (defun foo-1 (x y z) ...

** Functions 'region-to-string' and 'region-around-match' removed.

These functions were made for compatibility with Gosling Emacs, but it
turns out to be undesirable to use them in GNU Emacs because they use
the mark.  They have been eliminated from Emacs proper, but are
present in mlsupport.el for the sake of converted mocklisp programs.

If you were using 'region-to-string', you should instead use
'buffer-substring'; then you can pass the bounds as arguments and
can avoid setting the mark.

If you were using 'region-around-match', you can use instead
the two functions 'match-beginning' and 'match-end'.  These give
you one bound at a time, as a numeric value, without changing
point or the mark.

** Function 'function-type' removed.

This just appeared not to be very useful.  It can easily be written in
Lisp if you happen to want it.  Just use 'symbol-function' to get the
function definition of a symbol, and look at its data type or its car
if it is a list.

** Variable 'buffer-number' removed.

You can still use the function 'buffer-number' to find out
a buffer's unique number (assigned in order of creation).

** Variable 'executing-macro' renamed 'executing-kbd-macro'.

This variable is the currently executing keyboard macro, as
a string, or 'nil' when no keyboard macro is being executed.

** Loading term/$TERM.

The library term/$TERM (where $TERM get replaced by your terminal
type), which is done by Emacs automatically when it starts up, now
happens after the user's .emacs file is loaded.

In previous versions of Emacs, these files had names of the form
term-$TERM; thus, for example, term-vt100.el, but now they live
in a special subdirectory named term, and have names like
term/vt100.el.

** 'command-history' format changed.

The elements of this list are now Lisp expressions which can
be evaluated directly to repeat a command.

** Unused editing commands removed.

The functions 'forward-to-word', 'backward-to-word',
'upcase-char', 'mark-beginning-of-buffer' and 'mark-end-of-buffer'
have been removed.  Their definitions can be found in file
lisp/unused.el if you need them.


* Upward Compatible Lisp Programming Changes in Emacs 17

** You can now continue after errors and quits.

When the debugger is entered because of a C-g, due to
a non-'nil' value of 'debug-on-quit', the 'c' command in the debugger
resumes execution of the code that was running when the quit happened.
Use the 'q' command to go ahead and quit.

The same applies to some kinds of errors, but not all.  Errors
signaled with the Lisp function 'signal' can be continued; the 'c'
command causes 'signal' to return.  The 'r' command causes 'signal' to
return the value you specify.  The 'c' command is equivalent to 'r'
with the value 'nil'.

For a 'wrong-type-argument' error, the value returned with the 'r'
command is used in place of the invalid argument.  If this new value
is not valid, another error occurs.

Errors signaled with the function 'error' cannot be continued.
If you try to continue, the error just happens again.

** 'dot' renamed 'point'.

The word 'dot' has been replaced with 'point' in all
function and variable names, including:

  'point', 'point-min', 'point-max',
  'point-marker', 'point-min-marker', 'point-max-marker',
  'window-point', 'set-window-point',
  'point-to-register', 'register-to-point',
  'exchange-point-and-mark'.

The old names are still supported, for now.

** 'string-match' records position of end of match.

After a successful call to 'string-match', `(match-end 0)' will
return the index in the string of the first character after the match.
Also, 'match-begin' and 'match-end' with nonzero arguments can be
used to find the indices of beginnings and ends of substrings matched
by subpatterns surrounded by parentheses.

** New function 'insert-before-markers'.

This function is just like 'insert' except in the handling of any
relocatable markers that are located at the point of insertion.
With 'insert', such markers end up pointing before the inserted text.
With 'insert-before-markers', they end up pointing after the inserted
text.

** New function 'copy-alist'.

This function takes one argument, a list, and makes a disjoint copy
of the alist structure.  The list itself is copied, and each element
that is a cons cell is copied, but the cars and cdrs of elements
remain shared with the original argument.

This is what it takes to get two alists disjoint enough that changes
in one do not change the result of 'assq' on the other.

** New function 'copy-keymap'.

This function takes a keymap as argument and returns a new keymap
containing initially the same bindings.  Rebindings in either one of
them will not alter the bindings in the other.

** New function 'copy-syntax-table'.

This function takes a syntax table as argument and returns a new
syntax table containing initially the same syntax settings.  Changes
in either one of them will not alter the other.

** Randomizing the 'random' numbers.

`(random t)' causes the random number generator's seed to be set
based on the current time and Emacs's process id.

** Third argument to 'modify-syntax-entry'.

The optional third argument to 'modify-syntax-entry', if specified
should be a syntax table.  The modification is made in that syntax table
rather than in the current syntax table.

** New function 'run-hooks'.

This function takes any number of symbols as arguments.
It processes the symbols in order.  For each symbol which
has a value (as a variable) that is non-nil, the value is
called as a function, with no arguments.

This is useful in major mode commands.

** Second arg to 'switch-to-buffer'.

If this function is given a non-'nil' second argument, then the
selection being done is not recorded on the selection history.
The buffer's position in the history remains unchanged.  This
feature is used by the view commands, so that the selection history
after exiting from viewing is the same as it was before.

** Second arg to 'display-buffer' and 'pop-to-buffer'.

These two functions both accept an optional second argument which
defaults to 'nil'.  If the argument is not 'nil', it means that
another window (not the selected one) must be found or created to
display the specified buffer in, even if it is already shown in
the selected window.

This feature is used by 'switch-to-buffer-other-window'.

** New variable 'completion-ignore-case'.

If this variable is non-'nil', completion allows strings
in different cases to be considered matching.  The global value
is 'nil'

This variable exists for the sake of commands that are completing
an argument in which case is not significant.  It is possible
to change the value globally, but you might not like the consequences
in the many situations (buffer names, command names, file names)
where case makes a difference.

** Major modes related to Text mode call text-mode-hook, then their own hooks.

For example, turning on Outline mode first calls the value of
'text-mode-hook' as a function, if it exists and is non-'nil',
and then does likewise for the variable 'outline-mode-hook'.

** Defining new command line switches.

You can define a new command line switch in your .emacs file
by putting elements on the value of 'command-switch-alist'.
Each element of this list should look like
      (SWITCHSTRING . FUNCTION)
where SWITCHSTRING is a string containing the switch to be
defined, such as "-foo", and FUNCTION is a function to be called
if such an argument is found in the command line.  FUNCTION
receives the command line argument, a string, as its argument.

To implement a switch that uses up one or more following arguments,
use the fact that the remaining command line arguments are kept
as a list in the variable 'command-line-args'.  FUNCTION can
examine this variable, and do
    (setq command-line-args (cdr command-line-args)
to "use up" an argument.

** New variable 'load-in-progress'.

This variable is non-'nil' when a file of Lisp code is being read
and executed by 'load'.

** New variable 'print-length'.

The value of this variable is normally 'nil'.  It may instead be
a number; in that case, when a list is printed by 'prin1' or
'princ' only that many initial elements are printed; the rest are
replaced by `...'.

** New variable 'find-file-not-found-hook'.

If 'find-file' or any of its variants is used on a nonexistent file,
the value of 'find-file-not-found-hook' is called (if it is not 'nil')
with no arguments, after creating an empty buffer.  The file's name
can be found as the value of 'buffer-file-name'.

** Processes without buffers.

In the function 'start-process', you can now specify 'nil' as
the process's buffer.  You can also set a process's buffer to 'nil'
using 'set-process-buffer'.

The reason you might want to do this is to prevent the process
from being killed because any particular buffer is killed.
When a process has a buffer, killing that buffer kills the
process too.

When a process has no buffer, its output is lost unless it has a
filter, and no indication of its being stopped or killed is given
unless it has a sentinel.

** New function 'user-variable-p'.  'v' arg prompting changed.

This function takes a symbol as argument and returns 't' if
the symbol is defined as a user option variable.  This means
that it has a 'variable-documentation' property whose value is
a string starting with `*'.

Code 'v' in an interactive arg reading string now accepts
user variables only, and completion is limited to the space of
user variables.

The function 'read-variable' also now accepts and completes
over user variables only.

** CBREAK mode input is the default in Unix 4.3 bsd.

In Berkeley 4.3 Unix, there are sufficient features for Emacs to
work fully correctly using CBREAK mode and not using SIGIO.
Therefore, this mode is the default when running under 4.3.
This mode corresponds to 'nil' as the first argument to
'set-input-mode'.  You can still select either mode by calling
that function.

** Information on memory usage.

The new variable 'data-bytes-used' contains the number
of bytes of impure space allocated in Emacs.
'data-bytes-free' contains the number of additional bytes
Emacs could allocate.  Note that space formerly allocated
and freed again still counts as 'used', since it is still
in Emacs's address space.

** No limit on size of output from 'format'.

The string output from 'format' used to be truncated to
100 characters in length.  Now it can have any length.

** New errors 'void-variable' and 'void-function' replace 'void-symbol'.

This change makes it possible to have error messages that
clearly distinguish undefined variables from undefined functions.
It also allows 'condition-case' to handle one case without the other.

** 'replace-match' handling of `\'.

In 'replace-match', when the replacement is not literal,
`\' in the replacement string is always treated as an
escape marker.  The only two special `\' constructs
are `\&' and `\DIGIT', so `\' followed by anything other than
`&' or a digit has no effect.  `\\' is necessary to include
a `\' in the replacement text.

This level of escaping is comparable with what goes on in
a regular expression.  It is over and above the level of `\'
escaping that goes on when strings are read in Lisp syntax.

** New error 'invalid-regexp'.

A regexp search signals this type of error if the argument does
not meet the rules for regexp syntax.

** 'kill-emacs' with argument.

If the argument is a number, it is returned as the exit status code
of the Emacs process.  If the argument is a string, its contents
are stuffed as pending terminal input, to be read by another program
after Emacs is dead.

** New fifth argument to 'subst-char-in-region'.

This argument is optional and defaults to 'nil'.  If it is not 'nil',
then the substitutions made by this function are not recorded
in the Undo mechanism.

This feature should be used with great care.  It is now used
by Outline mode to make lines visible or invisible.

** ` *Backtrace*' buffer renamed to `*Backtrace*'.

As a result, you can now reselect this buffer easily if you switch to
another while in the debugger.

Exiting from the debugger kills the `*Backtrace*' buffer, so you will
not try to give commands in it when no longer really in the debugger.

** New function 'switch-to-buffer-other-window'.

This is the new primitive to select a specified buffer (the
argument) in another window.  It is not quite the same as
'pop-to-buffer', because it is guaranteed to create another
window (assuming there is room on the screen) so that it can
leave the current window's old buffer displayed as well.

All functions to select a buffer in another window should
do so by calling this new function.

** New variable 'minibuffer-help-form'.

At entry to the minibuffer, the variable 'help-form' is bound
to the value of 'minibuffer-help-form'.

'help-form' is expected at all times to contain either 'nil'
or an expression to be executed when C-h is typed (overriding
the definition of C-h as a command).  'minibuffer-help-form'
can be used to provide a different default way of handling
C-h while in the minibuffer.

** New \{...} documentation construct.

It is now possible to set up the documentation string for
a major mode in such a way that it always describes the contents
of the major mode's keymap, as it has been customized.
To do this, include in the documentation string the characters `\{'
followed by the name of the variable containing the keymap,
terminated with `}'.  (The `\' at the beginning probably needs to
be quoted with a second `\', to include it in the doc string.)
This construct is normally used on a line by itself, with no blank
lines before or after.

For example, the documentation string for the function 'c-mode' contains
    ...
    Paragraphs are separated by blank lines only.
    Delete converts tabs to spaces as it moves back.
    \\{c-mode-map}
    Variables controlling indentation style:
    ...

** New character syntax class "punctuation".

Punctuation characters behave like whitespace in word and
list parsing, but can be distinguished in regexps and in the
function 'char-syntax'.  Punctuation syntax is represented by
a period in 'modify-syntax-entry'.

** 'auto-mode-alist' no longer needs entries for backup-file names,

Backup suffixes of all kinds are now stripped from a file's name
before searching 'auto-mode-alist'.



* Changes in Emacs 16

** No special code for Ambassadors, VT-100's and Concept-100's.

Emacs now controls these terminals based on the termcap entry, like
all other terminals.  Formerly it did not refer to the termcap entries
for those terminal types, and often the termcap entries for those
terminals are wrong or inadequate.  If you experience worse behavior
on these terminals than in version 15, you can probably correct it by
fixing up the termcap entry.  See ./TERMS for more info.

See ./TERMS in any case if you find that some terminal does not work
right with Emacs now.

** Minibuffer default completion character is TAB (and not ESC).

So that ESC can be used in minibuffer for more useful prefix commands.

** C-z suspends Emacs in all modes.

Formerly, C-z was redefined for other purposes by certain modes,
such as Buffer Menu mode.  Now other keys are used for those purposes,
to keep the meaning of C-z uniform.

** C-x ESC (repeat-complex-command) allows editing the command it repeats.

Instead of asking for confirmation to re-execute a command from the
command history, the command is placed, in its Lisp form, into the
minibuffer for editing.  You can confirm by typing RETURN, change some
arguments and then confirm, or abort with C-g.

** Incremental search does less redisplay on slow terminals.

If the terminal baud rate is <= the value of 'isearch-slow-speed',
incremental searching outside the text on the screen creates
a single-line window and uses that to display the line on which
a match has been found.  Exiting or quitting the search restores
the previous window configuration and redisplays the window you
were searching in.

The initial value of 'isearch-slow-speed' is 1200.

This feature is courtesy of crl@purdue.

** Recursive minibuffers not allowed.

If the minibuffer window is selected, most commands that would
use the minibuffer gets an error instead.  (Specific commands
may override this feature and therefore still be allowed.)

Strictly speaking, recursive entry to the minibuffer is still
possible, because you can switch to another window after
entering the minibuffer, and then minibuffer-using commands
are allowed.  This is still allowed by a deliberate decision:
if you know enough to switch windows while in the minibuffer,
you can probably understand recursive minibuffers.

This may be overridden by binding the variable
'enable-recursive-minibuffers' to t.

** New major mode Emacs-Lisp mode, for editing Lisp code to run in Emacs.

The mode in which emacs lisp files is edited is now called 'emacs-lisp-mode'
and is distinct from 'lisp-mode'.  The latter is intended for use with
lisps external to Emacs.

The hook which is funcalled (if non-nil) on entry to elisp-mode is now
called 'emacs-lisp-mode-hook'.  A consequence of this changes is that
.emacs init files which set the value of 'lisp-mode-hook' may need to be
changed to use the new names.

** Correct matching of parentheses is checked on insertion.

When you insert a close-paren, the matching open-paren
is checked for validity.  The close paren must be the kind
of close-paren that the open-paren says it should match.
Otherwise, a warning message is printed.  close-paren immediately
preceded by quoting backslash syntax character is not matched.

This feature was originally written by shane@mit-ajax.

** M-x list-command-history
** M-x command-history-mode
** M-x electric-command-history

'list-command-history' displays forms from the command history subject
to user controlled filtering and limit on number of forms.  It leaves
the buffer in 'command-history-mode'.  M-x command-history-mode
recomputes the command history each time it is invoked via
'list-command-history'.  It is like Emacs-Lisp mode except that characters
don't insert themselves and provision is made for re-evaluating an
expression from the list.  'electric-command-history' pops up a type
out window with the command history displayed.  If the very next
character is Space, the window goes away and the previous window
configuration is restored.  Otherwise you can move around in the
history and select an expression for evaluation *inside* the buffer
which invoked 'electric-command-history'.  The original window
configuration is restored on exit unless the command selected changes
it.

** M-x edit-picture

Enters a temporary major mode (the previous major mode is remembered
and can is restored on exit) designed for editing pictures and tables.
Printing characters replace rather than insert themselves with motion
afterwards that is user controlled (you can specify any of the 8
compass directions).  Special commands for movement are provided.
Special commands for hacking tabs and tab stops are provided.  Special
commands for killing rectangles and overlaying them are provided.  See
the documentation of function  edit-picture  for more details.

Calls value of 'edit-picture-hook' on entry if non-nil.

** Stupid C-s/C-q `flow control' supported.

Do (set-input-mode nil t) to tell Emacs to use CBREAK mode and interpret
C-s and C-q as flow control commands.  (set-input-mode t nil) switches
back to interrupt-driven input.  (set-input-mode nil nil) uses CBREAK
mode but no `flow control'; this may make it easier to run Emacs under
certain debuggers that have trouble dealing with inferiors that use SIGIO.

CBREAK mode has certain inherent disadvantages, which are why it is
not the default:

     Meta-keys are ignored; CBREAK mode discards the 8th bit of
     input characters.

     Control-G as keyboard input discards buffered output,
     and therefore can cause incorrect screen updating.

The use of `flow control' has its own additional disadvantage: the
characters C-s and C-q are not available as editing commands.  You can
partially compensate for this by setting up a keyboard-translate-table
(see file ONEWS) that maps two other characters (such as C-^ and C-\) into
C-s and C-q.  Of course, C-^ and C-\ are commonly used as escape
characters in remote-terminal programs.  You really can't win except
by getting rid of this sort of `flow control.'

The configuration switch CBREAK_INPUT is now eliminated.
INTERRUPT_INPUT exists only to specify the default mode of operation;
#define it to make interrupt-driven input the default.

** Completion of directory names provides a slash.

If file name completion yields the name of a directory,
a slash is appended to it.

** Undo can clear modified-flag.

If you undo changes in a buffer back to a state in which the
buffer was not considered "modified", then it is labeled as
once again "unmodified".

** M-x run-lisp.

This command creates an inferior Lisp process whose input and output
appear in the Emacs buffer named `*lisp*'.  That buffer uses a major mode
called inferior-lisp-mode, which has many of the commands of lisp-mode
and those of shell-mode.  Calls the value of shell-mode-hook and
lisp-mode-hook, in that order, if non-nil.

Meanwhile, in lisp-mode, the command C-M-x is defined to
send the current defun as input to the `*lisp*' subprocess.

** Mode line says 'Narrow' when buffer is clipped.

If a buffer has a clipping restriction (made by 'narrow-to-region')
then its mode line contains the word 'Narrow' after the major and
minor modes.

** Mode line says 'Abbrev' when abbrev mode is on.

** 'add-change-log-entry' takes prefix argument

Giving a prefix argument makes it prompt for login name, full name,
and site name, with defaults.  Otherwise the defaults are used
with no confirmation.

** M-x view-buffer and M-x view-file

view-buffer selects the named buffer, view-file finds the named file; the
resulting buffer is placed into 'view-mode' (a recursive edit).  The normal
emacs commands are not available.  Instead a set of special commands is
provided which facilitate moving around in the buffer, searching and
scrolling by screenfuls.  Exiting view-mode returns to the buffer in which
the view-file or view-buffer command was given.
Type ? or h when viewing for a complete list of view commands.
Each calls value of 'view-hook' if non-nil on entry.

written by shane@mit-ajax.

** New key commands in dired.

'v' views (like more) the file on the current line.
`#' marks auto-save files for deletion.
`~' marks backup files for deletion.
'r' renames a file and updates the directory listing if the
file is renamed to same directory.
'c' copies a file and updates the directory listing if the file is
copied to the same directory.

** New function 'electric-buffer-list'.

This pops up a buffer describing the set of emacs buffers.
Immediately typing space makes the buffer list go away and returns
to the buffer and window which were previously selected.

Otherwise one may use the c-p and c-n commands to move around in the
buffer-list buffer and type Space or C-z to select the buffer on the
cursor's line.  There are a number of other commands which are the same
as those of buffer-menu-mode.

This is a useful thing to bind to c-x c-b in your `.emacs' file if the
rather non-standard 'electric' behavior of the buffer list suits your taste.
Type C-h after invoking electric-buffer-list for more information.

Calls value of 'electric-buffer-menu-mode-hook' if non-nil on entry.
Calls value of 'after-electric-buffer-menu' on exit (select) if non-nil.

** Changes in version 16 for mail reading and sending

*** sendmail prefix character is C-c (and not C-z).  New command C-c w.

For instance C-c C-c (or C-c C-s) sends mail now rather than C-z C-z.
C-c w inserts your 'signature' (contents of ~/.signature) at the end
of mail.

*** New feature in C-c y command in sending mail.

C-c y is the command to insert the message being replied to.
Normally it deletes most header fields and indents everything
by three spaces.

Now, C-c y does not delete header fields or indent.
C-c y with any other numeric argument does delete most header
fields, but indents by the amount specified in the argument.

*** C-r command in Rmail edits current message.

It does this by switching to a different major mode
which is nearly the same as Text mode.  The only difference
between it and text mode are the two command C-c and C-].
C-c is defined to switch back to Rmail mode, and C-]
is defined to restore the original contents of the message
and then switch back to Rmail mode.

C-c and C-] are the only ways "back into Rmail", but you
can switch to other buffers and edit them as usual.
C-r in Rmail changes only the handling of the Rmail buffer.

*** Rmail command 't' toggles header display.

Normally Rmail reformats messages to hide most header fields.
't' switches to display of all the header fields of the
current message, as long as it remains current.
Another 't' switches back to the usual display.

*** Rmail command '>' goes to the last message.

*** Rmail commands 'a' and 'k' set message attributes.
'a' adds an attribute and 'k' removes one.  You specify
the attribute by name.  You can specify either a built-in
flag such as "deleted" or "filed", or a user-defined keyword
(anything not recognized as built-in).

*** Rmail commands 'l' and 'L' summarize by attributes.

These commands create a summary with one line per message,
like 'h', but they list only some of the messages.  You
specify which attribute (for 'l') or attributes (for 'L')
the messages should have.

*** Rmail can parse mmdf mail files.

*** Interface to MH mail system.

mh-e is a front end for GNU emacs and the MH mail system.  It
provides a friendly and convenient interface to the MH commands.

To read mail, invoke 'mh-rmail'.  This will inc new mail and display the
scan listing on the screen.  To see a summary of the mh-e commands,
type ?.  Help is available through the usual facilities.

To send mail, invoke 'mh-smail'.

mh-e requires a copy of MH.5 that has been compiled with the MHE
compiler switch.

From larus@berkeley.

** New hooks and parameters in version 16

*** New variable 'blink-matching-paren-distance'.

This is the maximum number of characters to search for
an open-paren to match an inserted close-paren.
The matching open-paren is shown and checked if it is found
within this distance.

'nil' means search all the way to the beginning of the buffer.
In this case, a warning message is printed if no matching
open-paren is found.

This feature was originally written by shane@mit-ajax.

*** New variable 'find-file-run-dired'

If nil, find-file will report an error if an attempt to visit a
directory is detected; otherwise, it runs dired on that directory.
The default is t.

*** Variable 'dired-listing-switches' holds switches given to 'ls' by dired.

The value should be a string containing '-' followed by letters.
The letter 'l' had better be included and letter 'F' had better be excluded!
The default is "-al".

This feature was originally written by shane@mit-ajax.

*** New variable 'display-time-day-and-date'.

If this variable is set non-'nil', the function M-x display-time
displays the day and date, as well as the time.

*** New parameter 'c-continued-statement-indent'.

This controls the extra indentation given to a line
that continues a C statement started on the previous line.
By default it is 2, which is why you would see

	if (foo)
	  bar ();


*** Changed meaning of 'c-indent-level'.

The value of 'c-brace-offset' used to be
subtracted from the value of 'c-indent-level' whenever
that value was used.  Now it is not.

As a result, 'c-indent-level' is now the offset of
statements within a block, relative to the line containing
the open-brace that starts the block.

*** 'turn-on-auto-fill' is useful value for 'text-mode-hook'.

    (setq text-mode-hook 'turn-on-auto-fill)

is all you have to do to make sure Auto Fill mode is turned
on whenever you enter Text mode.

*** Parameter explicit-shell-file-name for M-x shell.

This variable, if non-nil, specifies the file name to use
for the shell to run if you do M-x shell.

Changes in version 16 affecting Lisp programming:

*** Documentation strings adapt to customization.

Often the documentation string for a command wants to mention
another command.  Simply stating the other command as a
character sequence has a disadvantage: if the user customizes
Emacs by moving that function to a different command, the
cross reference in the documentation becomes wrong.

A new feature allows you to write the documentation string
using a function name, and the command to run that function
is looked up when the documentation is printed.

If a documentation string contains `\[' (two characters) then
the following text, up to the next `]', is taken as a function name.
Instead of printing that function name, the command that runs it is printed.
(M-x is used to construct a command if no shorter one exists.)

For example, instead of putting 'C-n' in a documentation string
to refer to the 'next-line' command, put in `\[next-line]'.  (In practice
you will need to quote the backslash with another backslash,
due to the syntax for strings in Lisp and C.)

To include the literal characters `\[' in a documentation string,
precede them with `\='.  To include the characters `\=', precede
them with `\='.  For example, "\\=\\= is the way to quote \\=\\["
will come out as `\= is the way to quote \['.

The new function 'substitute-command-keys' takes a string possibly
containing \[...] constructs and replaces those constructs with
the key sequences they currently stand for.

*** Primitives 'find-line-comment' and 'find-line-comment-body' flushed.

Search for the value of 'comment-start-skip' if you want to find
whether and where a line has a comment.

*** New function 'auto-save-file-name-p'

Should return non-'nil' if given a string which is the name of an
auto-save file (sans directory name).  If you redefine
'make-auto-save-file-name', you should redefine this accordingly.  By
default, this function returns 't' for filenames beginning with
character `#'.

*** The value of 'exec-directory' now ends in a slash.

This is to be compatible with most directory names in GNU Emacs.

*** Dribble files and termscript files.

'open-dribble-file' opens a dribble file.  When a
dribble file is open, every character Emacs reads from the terminal is
written to the dribble file.

'open-termscript' opens a termscript file.  When a
termscript file is open, all characters sent to the terminal by Emacs
are also written in the termscript file.

The two of these together are very useful for debugging Emacs problems
in redisplay.

*** Upper case command characters by default are same as lower case.

If a character in a command is an upper case letter, and is not defined,
Emacs uses the definition of the corresponding lower case letter.
For example, if C-x U is not directly undefined, it is treated as
a synonym for C-x u (undo).

*** Undefined function errors versus undefined variable errors.

Void-symbol errors now say 'boundp' if the symbol's value was void
or 'fboundp' if the function definition was void.

*** New function 'bury-buffer'.

The new function 'bury-buffer' takes one argument, a buffer object,
and puts that buffer at the end of the internal list of buffers.
So it is the least preferred candidate for use as the default value
of C-x b, or for 'other-buffer' to return.

*** Already-displayed buffers have low priority for display.

When a buffer is chosen automatically for display, or to be the
default in C-x b, buffers already displayed in windows have lower
priority than buffers not currently visible.

*** 'set-window-start' accepts a third argument NOFORCE.

This argument, if non-nil, prevents the window's force_start flag
from being set.  Setting the force_start flag causes the next
redisplay to insist on starting display at the specified starting
point, even if dot must be moved to get it onto the screen.

*** New function 'send-string-to-terminal'.

This function takes one argument, a string, and outputs its contents
to the terminal exactly as specified: control characters, escape
sequences, and all.

*** Keypad put in command mode.

The terminal's keypad is now put into command mode, as opposed to
numeric mode, while Emacs is running.  This is done by means of the
termcap 'ks' and 'ke' strings.

*** New function 'generate-new-buffer'

This function takes a string as an argument NAME and looks for a
creates and returns a buffer called NAME if one did not already exist.
Otherwise, it successively tries appending suffixes of the form "<1>",
"<2>" etc to NAME until it creates a string which does not name an
existing buffer.  A new buffer with that name is the created and returned.

*** New function 'prin1-to-string'
This function takes one argument, a lisp object, and returns a string
containing that object's printed representation, such as 'prin1'
would output.

*** New function 'read-from-minibuffer'
Lets you supply a prompt, initial-contents, a keymap, and specify
whether the result should be interpreted as a string or a lisp object.

Old functions 'read-minibuffer', 'eval-minibuffer', 'read-string' all
take second optional string argument which is initial contents of
minibuffer.

*** minibuffer variable names changed (names of keymaps)

'minibuf-local-map' -> 'minibuffer-local-map'
'minibuf-local-ns-map' -> 'minibuffer-local-ns-map'
'minibuf-local-completion-map' -> 'minibuffer-local-completion-map'
'minibuf-local-must-match-map' -> 'minibuffer-local-must-match-map'

** Changes in version 16 affecting configuring and building Emacs

*** Configuration switch VT100_INVERSE eliminated.

You can control the use of inverse video on any terminal by setting
the variable 'inverse-video', or by changing the termcap entry.  If
you like, set 'inverse-video' in your `.emacs' file based on
examination of (getenv "TERM").

*** New switch '-batch' makes Emacs run noninteractively.

If the switch '-batch' is used, Emacs treats its standard output
and input like ordinary files (even if they are a terminal).
It does not display buffers or windows; the only output to standard output
is what would appear as messages in the echo area, and each
message is followed by a newline.

The terminal modes are not changed, so that C-z and C-c retain
their normal Unix meanings.  Emacs does still read commands from
the terminal, but the idea of '-batch' is that you use it with
other command line arguments that tell Emacs a complete task to perform,
including killing itself.  '-kill' used as the last argument is a good
way to accomplish this.

The Lisp variable 'noninteractive' is now defined, to be 'nil'
except when '-batch' has been specified.

*** Emacs can be built with output redirected to a file.

This is because -batch (see above) is now used in building Emacs.



* 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 initially 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 directories
 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 unmetized
 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
 designs.  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 that 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' 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 and 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 like 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' 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, use 'debug-on-entry'.  To cancel this,
 use 'cancel-debug-on-entry'.  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 mycadr (arg) (list 'car (list 'cdr arg)))
 and then the expression
    (mycadr 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 heading 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 diredded 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.

*** New function 'pos-visible-in-window-p'

    (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'.

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.



----------------------------------------------------------------------
This file is part of GNU Emacs.

GNU Emacs is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

GNU Emacs is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with GNU Emacs.  If not, see <https://www.gnu.org/licenses/>.


Local variables:
mode: outline
end: