procyberian/emacs
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Rewrite Antinews in ELisp manual for Emacs 29

Browse source 
* doc/lispref/anti.texi (Antinews): Rewrite for Emacs 29.
* doc/lispref/elisp.texi (Top): Update the top-level menu for
Antinews.
This commit is contained in:
Eli Zaretskii 2023-01-01 19:37:22 +02:00
parent f12f72b0e0
commit 7b0b17df67
1 changed files with 116 additions and 113 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

229
doc/lispref/anti.texi
View file

@ -6,179 +6,182 @@
@c This node must have no pointers.
@node Antinews
@appendix Emacs 27 Antinews
@appendix Emacs 28 Antinews
@c Update the elisp.texi Antinews menu entry with the above version number.
For those users who live backwards in time, here is information about
downgrading to Emacs version 27.2. We hope you will enjoy the greater
downgrading to Emacs version 28.2. We hope you will enjoy the greater
simplicity that results from the absence of many @w{Emacs
@value{EMACSVER}} features.
@itemize @bullet
@item
The annoying @code{lexical-binding} local variable now heeds the
value of @code{enable-local-variables}: if it's @code{nil}, the
@code{lexical-binding} cookie is ignored. We are working hard on
removing the lexical-binding support in some past Emacs version, and
this small step advances us back to that change.
The implementation of overlays is back to its simple, time-proven
storage in a pair of linear linked lists centered around some buffer
position. No more fancy interval trees and suchlikes. Lisp programs
that use overlays once again need to recenter overlays around the
buffer position of interest, and display-related features should again
make sure they don't use too many overlays in a buffer, lest redisplay
will be too slow.
@item
The @code{load-dangerous-libraries} variable is not obsolete, as it
must be used to allow loading Lisp compiled by XEmacs, which will
become more and more important as you move back in time.
Several functions stopped the annoying conversion of quotes and key
sequences by no longer calling @code{substitute-command-keys}. One
prominent example is @code{format-prompt} and all its many callers.
This makes the strings they produce much more predictable, returning
to you, the Lisp programmer, control on which punctuation characters
will appear in the text presented to the users. For similar reasons,
the @code{substitute-quotes} function was deleted.
@item
The optional @var{modes} argument of @code{interactive} is not
supported, and every command is deemed applicable to any major mode.
We believe this makes the life of Lisp programmers much simpler, as
there's now no need to tag commands with the modes where they make
sense.
The venerable @code{buffer-modified-p} function again reliably returns
either @code{nil} or @code{t}, not any other confusing values.
@item
Shorthands for Lisp symbols have been removed, which makes loading
Lisp files and handling Lisp symbols much simpler and more efficient.
This is important for decent performance on slower CPUs as you move
back in time.
The support for @samp{medium} weight of fonts was dropped. Emacs now
considers @samp{medium} and @samp{regular} weights to be the same. We
believe this will simplify your font setup, since there's no longer a
need to worry about fonts that support @samp{regular} weight, but not
the @samp{medium} one, or vice versa: either one will do!
@item
To reduce the amount of code in Emacs related to unimportant features,
we've removed the variables @code{global-minor-modes} and
@code{local-minor-modes}. If your Lisp program needs to determine
whether some minor mode is in effect, it will have to test explicitly
for every mode. We don't expect anyone to miss those fancy variables.
we've removed the function @code{compiled-function-p}. Lisp programs
are expected to test explicitly for the relevant types of function
objects: built-in, byte-compiled, and natively-compiled. For the same
reasons we deleted the functions @code{pos-bol}, @code{pos-eol},
@code{file-attribute-file-identifier}, and quite a few others. We
don't expect anyone to miss those fancy functions.
@item
The default preference for servicing sub-processes that produce output
at a high rate, and the associated variable
@code{process-prioritize-lower-fds}, have been removed. Moving back
in time means fewer and fewer programs can produce such high-rate
output, so this features becomes just useless crud.
The timeout used by @code{x-show-tip} can no longer be specified by
Lisp programs; it is hard-coded in the function. This will lead to a
simpler, easier maintained code, and no one should want to control the
timeout after which the tip pops down.
@item
The encodings that are variants of EBCDIC were removed. This includes
@code{ibm256}, @code{ibm273}, and others---variants of the EBCDIC
encoding tailored for some Japanese and European locales. You won't
need those where you are going.
The macro @code{setopt} was deleted; use @code{customize-variable}
instead, or invoke the @code{:set} function from Lisp.
@item
The ``Bindat type expression'' description language has been removed,
as the existing data layout specifications are perfectly suited for
this job.
We removed the @code{lisp-directory} variable, as the value can be
easily deduced from other similar variables, like
@code{installation-directory} and @code{source-directory}, each one
when it's relevant.
@item
To simplify code and reduce complexity, we removed the capability of
specifying the success handler in @code{condition-case} via the
@code{:success} keyword. If you really need this feature (why would
you?), you can always write some simple Lisp that has the same effect.
To simplify code and reduce complexity, we deleted the functions
@code{get-display-property} and @code{add-display-text-property}; use
the generic @code{get-text-property} and @code{put-text-property}
instead.
@item
Emacs modules can no longer provide interactive functions, or install
finalizers, nor open channels to existing pipe sub-processes. All
this is extra ballast, especially since we plan on removing modules in
some past Emacs version. The @code{make_unibyte_string} module API
was removed for the same reason.
Support for pinch input events and for modern drag-and-drop
functionality on X was dropped. As you move back in time, these
facilities will become less and less important, and will soon enough
disappear, so there's no reason to keep them in Emacs.
@item
To keep Emacs clean and elegant, we've removed the
@code{print-integers-as-characters} option. Recognizing characters by
their decimal codes is a basic requirement for Emacs Lisp programmers,
and with the expected decrease in use of Unicode characters, this will
be soon limited to ASCII only: surely something you all can master!
To keep Emacs clean and elegant, we've removed the @file{textsec.el}
library, with its facilities for checking whether some text is
``suspicious''. We consider our users smart enough to detect
maliciously modified text by just looking at it or by moving the
cursor across it, and the whole idea that someone would wish to
deliberately deceive Emacs users ridiculous and unworthy of
complicating our elegant text-processing and display capabilities.
@item
The optional @var{count} argument of the @code{directory-files}
function has been removed. Extracting the first @var{n} members from
the full list is trivial, so this is a significant simplification for
an insignificant cost.
The functions @code{keymap-set}, @code{keymap-global-set},
@code{keymap-local-set}, @code{keymap-substitute},
@code{keymap-lookup}, and some others were deleted. We have found the
traditional @code{define-key}, @code{global-set-key},
@code{local-set-key}, @code{substitute-key-definition}, and
@code{key-binding} more than enough, and their minor inconsistencies
in the syntax of keys they accept a source of endless fun in Emacs
Lisp programming. Why make Emacs programming a dull place? For the
same reasons we deleted @code{key-valid-p}, since we consider the
permissive nature of @code{kbd} more in the spirit of Emacs Lisp.
@item
Functions that create sub-processes and network connections no longer
accept the @code{:coding} argument; use
@code{set-process-coding-system} or bind
@code{coding-system-for-read/write} instead: again, a significant
reduction in Emacs complexity for little or no cost.
Yanking of anything but plain text from other applications becomes
more and more an unnecessary feature as you move back in time, so we
dropped support for pasting media like HTML and images via the
clipboard. If you @i{really} need to yank those into an Emacs buffer,
you can go via a disk file.
@item
We deleted from the macros @code{define-derived-mode} and
@code{define-minor-mode} the code which allowed using the
@code{:interactive} argument. The possibility of marking a mode
non-interactive makes very little sense,
We removed unnecessary functions @code{string-pixel-width} and
@code{string-glyph-split}, as we consider it inappropriate for Lisp
programs to do display layout calculations, where these functions come
in handy. Display is for the display engine, written in C, and should
stay there!
@item
The possibility of having links to man pages in doc strings has been
removed. Use plain text instead, if you need such references.
Various new Xwidget functions, such as
@code{xwidget-perform-lispy-event}, @code{xwidget-webkit-load-html},
and @code{xwidget-webkit-back-forward-list}, were deleted as part of
our continuing effort to gradually delete the entire Xwidget
functionality in some previous release of Emacs.
@item
Temporary buffers are no longer exempt from running any buffer-related
hooks. Programs that don't want such hooks in some buffer can always
disable it locally, whereas making that simpler complicates Emacs for
no good reason.
Setting the @code{:stderr} property of a process in a
@code{make-process} call once again forces the process's connection to
use pipes, not ptys, for all the standard streams --- a considerable
simplification of this complex interface.
@item
To keep the amount of Lisp functions from growing out of control, we
deleted @code{string-equal-ignore-case}. Use @code{compare-strings}
instead.
Several features that complicated the byte compiler have been removed:
@itemize @minus
@item
The checks for missing declarations of dynamic variables. This will
continue making less and less sense as we move away of lexical-binding
support.
The warnings about quoting mistakes in documentation strings. You are
expected to find such mistakes yourself, by eyeballing the resulting
@file{*Help*} buffer display.
@item
The ability of compiling symlinked @file{*.el} files, which is really
gross: copy the files instead.
@item
The warnings about too-wide doc strings---that is just a nuisance, as
the programmers should be trusted to know what they are doing.
@end itemize
@item
We deleted several features of the @code{pcase} macro, in accordance
with our general plan to remove @code{pcase} from Emacs:
@itemize @minus
@item
The @code{cl-type} pattern.
@item
the @code{pcase-setq} macro.
@item
The @code{pcase-compile-patterns} function.
The warnings about malformed @code{defcustom} types, like
double-quoting symbols in @code{choice} lists.
@end itemize
@item
Some of the keywords used in Edebug specification lists were deemed to
be of little use, and were therefore removed: @code{&interpose},
@code{&error}, and @code{&name}. The long-term plane is for Emacs to
drop Edebug entirely, leaving only the trusted Lisp debugger, and we
continue working according to that plan.
The macro @code{with-buffer-unmodified-if-unchanged} was deleted.
Lisp programs that need to leave the buffer unmodified in these cases
can always compare the text before and after the modifications.
@item
The function @code{object-intervals} was dropped, as a Lisp program
can easily collect the intervals of a buffer or a string by iterating
through them one by one.
The functions @code{string-edit} and @code{read-string-from-buffer}
were removed, as we consider the fun of programming them anew every
time an important part of the education of each Emacs Lisp developer.
@item
We decided that the @code{require-theme} function is an unnecessary
complication, so we deleted it. Lisp programs can easily search along
@code{custom-theme-load-path} instead.
We deleted the function @code{readablep} and the related variable
@code{print-unreadable-function}, since no one is supposed to want to
print unreadable Lisp objects.
@item
The convenience functions @code{length<}, @code{length>}, and
@code{length=} were removed, as using @code{length} followed by a
comparison should be good enough for everyone, especially considering
that the typical length of a list keeps going down as you move back
through time.
The facility for storing multisession variables was deleted as an
unnecessary complication. With it are gone @code{multisession-value},
@code{define-multisession-variable}, and
@code{list-multisession-values}.
@item
The variable @code{current-minibuffer-command} is no longer available,
as we found little justification for keeping it.
The support for the @code{cursor-face} text property was dropped. We
consider the rest of the faces adequate for supporting this
functionality.
@item
The function @code{tooltip-show} dropped support for optional face
arguments @code{text-face} and @code{default-face} that allow fancy
control of the face of the tip text and top frame colors. We decided
that tooltips should all look the same, to prevent user confusion.
@item
As part of the ongoing quest for simplicity, many other functions and
variables have been eliminated. Other functions and variables, that
were declared obsolete since Emacs 23, have been added back, in
preparation for releasing Emacs 23 in some distant past.
were declared obsolete since Emacs 24, have been added back, in
preparation for releasing Emacs 24 in some distant past.
@end itemize