* lisp/emacs-lisp/bytecomp.el (byte-compile-associative):
Translate numerical identity expressions, such as (+ x) and (* x),
into (* x 1) since the previous translation (+ x 0) gets it wrong
for x = -0.0.
* test/lisp/emacs-lisp/bytecomp-tests.el
(byte-opt-testsuite-arith-data): Add test cases.
* lisp/emacs-lisp/autoload.el (batch-update-autoloads--summary):
Output " ..." at the end of the non-concluding lines to signify
that the output continues.
* lisp/Makefile.in ($(lisp)/loaddefs.el): Don't output the
directories here.
* lisp/emacs-lisp/autoload.el (batch-update-autoloads--summary):
New function.
(batch-update-autoloads): Use it to output the directories we're
scraping.
This changes the compilation output from:
Directories for loaddefs: . ./calc ./calendar ./cedet ./cedet/ede
./cedet/semantic ./cedet/semantic/analyze ./cedet/semantic/bovine
./cedet/semantic/decorate ./cedet/semantic/symref ./cedet/semantic/wisent
./cedet/srecode ./emacs-lisp ./emulation ./erc ./eshell ./gnus ./image
./international ./language ./leim ./leim/ja-dic ./leim/quail ./mail ./mh-e
./net ./nxml ./org ./play ./progmodes ./textmodes ./url ./vc
(but all on one long line)
To:
SCRAPE . ./calc ./calendar ./cedet ./cedet/ede ./cedet/semantic
SCRAPE ./cedet/semantic/analyze ./cedet/semantic/bovine
SCRAPE ./cedet/semantic/decorate ./cedet/semantic/symref
SCRAPE ./cedet/semantic/wisent ./cedet/srecode ./emacs-lisp ./emulation
SCRAPE ./erc ./eshell ./gnus ./image ./international ./language ./leim
SCRAPE ./leim/ja-dic ./leim/quail ./mail ./mh-e ./net ./nxml ./org ./play
SCRAPE ./progmodes ./textmodes ./url ./vc
Compilation output with very long lines can be mistaken for errors
when they scroll by fast in the compilation output. Making it look
more like normal informational output avoids this confusion.
* lisp/emacs-lisp/byte-run.el (byte-compile-info): New function
that's more flexible that replaces 'byte-compile-info-string' and
'byte-compile-info-message'.
(byte-compile-info-string): Make obsolete.
(byte-compile-info-message): Ditto.
* lisp/international/ja-dic-cnv.el (skkdic-convert-okuri-ari)
(skkdic-convert-postfix, skkdic-convert-prefix)
(skkdic-collect-okuri-nasi, skkdic-set-okuri-nasi):
* lisp/finder.el (finder-compile-keywords):
* lisp/cus-dep.el (custom-make-dependencies): Adjust callers to
use the new function.
When Edebug backtracks, it nevertheless generates definitions for the
non-matching branches, see Bug#41988 and Bug#42701. This should be
fixed eventually (probably by deferring the definition until a branch
is known to match), but for now add a band-aid to avoid these
duplicate symbols, at least for anonymous forms.
* lisp/emacs-lisp/edebug.el (edebug-make-enter-wrapper): Regenerate
anonymous names.
* test/lisp/emacs-lisp/edebug-tests.el
(edebug-tests-duplicate-symbol-backtrack): New regression test.
* lisp/emacs-lisp/cl-generic.el (cl-defgeneric): Include qualifiers in
Edebug symbol name.
* test/lisp/emacs-lisp/cl-generic-tests.el
(cl-defgeneric/edebug/method): Adapt unit test.
* lisp/emacs-lisp/cl-generic.el (cl-defgeneric): Work around Bug#42672
by uniquifying inline method names.
* test/lisp/emacs-lisp/cl-generic-tests.el
(cl-defgeneric/edebug/method): New regression test.
* lisp/emacs-lisp/edebug.el (edebug-match-:unique): Add a new
‘:unique’ specifier to generate unique names.
* lisp/emacs-lisp/cl-macs.el (cl-flet): Use it. This requires
inlining the ‘cl-defun’ specification.
* test/lisp/emacs-lisp/edebug-tests.el (edebug-tests-cl-flet): New
unit test.
* doc/lispref/edebug.texi (Specification List): Document new ‘:unique’
construct.
* lisp/emacs-lisp/edebug.el
(edebug-match-cl-generic-method-qualifier): Add matcher for
‘cl-defmethod’ qualifier.
* lisp/emacs-lisp/cl-generic.el (cl-defmethod): Use it.
* test/lisp/emacs-lisp/edebug-tests.el
(edebug-cl-defmethod-qualifier): New unit test.
(byte-optimize-form-code-walker): Use `byte-optimize-form` after
inlining, so optimizations are also applied to the top level call.
Simplify the code for `pure` functions using `byte-optimize-constant-args`.
(byte-optimize-all-constp): Remove, not used any more.
(byte-optimize-1+, byte-optimize-1-): Remove, they are redundant
with the `pure` annotation.
* lisp/comint.el (comint-read-noecho):
* lisp/emacs-lisp/edebug.el (edebug-all-defuns):
* lisp/man.el (manual-entry):
* lisp/vc/log-edit.el (vc-log-mode-map, vc-log-entry-mode): Declare
ancient backwards-compatibility aliases and functions obsolete. The
oldest in this list was added in 1992, and the most recent in 2004.
* lisp/net/telnet.el (telnet-initial-filter): Don't use
`comint-read-noecho'.
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker):
Remove clause for 'with-output-to-temp-buffer', since it is a
macro and will have been expanded before reaching this point.
Move clauses for 'lambda' and 'closure' to avoid splitting
a cond jump table.
Fixes: bug#42563
For some time, Eldoc has has some Elisp-specific code that shouldn't
live there, but in elisp-mode.el. This can be fixed in Emacs master,
but since ElDoc is distributed in GNU Elpa and is meant to work with
Emacs 26 and 27, this means that that elisp-specific code must still
be distributed with eldoc.el and kept up to date.
* lisp/emacs-lisp/eldoc.el (eldoc--eval-expression-setup): Ensure
implementation is compatible with Emacs < 28.
(Version): Bump to 1.8
Like others, this is an implementation detail that third parties could
be relying on. Better not remove it outright just now, since its
implementation is very simple anyway.
* lisp/emacs-lisp/eldoc.el (Version): Bump to 1.7.0
(eldoc-display-message-p): Bring back, but obsolete.
* lisp/emacs-lisp/comp.el (comp-ensure-native-compiler): New function.
(native-compile, batch-native-compile)
(batch-byte-native-compile-for-bootstrap, native-compile-async):
Make use of `comp-ensure-native-compiler'.
Turn (+ a b c) into (+ (+ a b) c), and do the same for - and *.
The 2-arg operations have their own bytecode which results in a 1.5×
speed-up. Furthermore, the transform enables other optimisations; for
example, (+ a 1 b) -> (+ (1+ a) b).
* lisp/emacs-lisp/byte-opt.el (byte-optimize-plus, byte-optimize-minus)
(byte-optimize-multiply): Transform (OP a b c) into (OP (OP a b) c).
Fixes: bug#42421
Do this conservatively for now: if the ElDoc helper buffer (as
returned by eldoc--doc-buffer) is visible and showing documentation
for the very same "situation" (as computed by the the new
eldoc--request-state helper), don't request that documentation from
sources again.
Before this change, not only was that request inefficient but if the
user invoked scroll-other-window to see more of the helper buffer,
that would eventually cause it to be reformatted and unexpectedly
recentered.
Later on, when a customizable list of documentation "sinks" is offered
to the user, say, something like eldoc-display-functions, this process
must be consolidated. In those circumstances, as soon as one of those
sinks signals that it doesn't have up-to-date documentation for the
state computed by eldoc--request-state, documentation will have to be
requested anew from eldoc-documentation-functions via
eldoc--invoke-strategy.
* lisp/emacs-lisp/eldoc.el (eldoc--request-docs-p): Rework from
eglot-display-message-p.
(eldoc--last-request-state): New variable.
(eldoc--request-state): New helper.
(eldoc--handle-docs): Memorize state of request in doc buffer.
(eldoc-print-current-symbol-info): Pass a token to
eldoc--request-docs-p.
(Version): Bump to 1.6.0
If the value is 'truncate-sym-name-if-fit andthe single docstring
doesn't fit in the echo area even when the symbol name is elided, that
step shouldn't be attempted. Also if the value is nil, really ensure
that only the first line is shown.
* lisp/emacs-lisp/eldoc.el (eldoc--handle-docs): Rework
'truncate-sym-name-if-fit case of eldoc-echo-area-use-multiline-p.
* lisp/emacs-lisp/text-property-search.el
(text-property-search-backward): Fix inconsistent behaviour of
S-TAB in eww (and other callers that use the NOT-CURRENT
behaviour) when there are adjacent elements
(bug#39239).
* lisp/emacs-lisp/comp.el (comp-limple-lock-keywords): New const.
(comp-limple-mode): New major mode.
(comp-log-to-buffer): Enable `comp-limple-mode' in the log buffer.
This fixes new occurrences of "Eldoc" since emacs-27.
* doc/lispref/modes.texi (Major Mode Conventions):
* etc/NEWS:
* lisp/descr-text.el (describe-char-eldoc):
* lisp/emacs-lisp/eldoc.el (eldoc-echo-area-use-multiline-p)
(eldoc-prefer-doc-buffer, eldoc--documentation-strategy-defcustom):
Consistently capitalize eldoc as ElDoc rather than Eldoc in
documentation and commentary.
* lisp/emacs-lisp/comp.el (comp-passes): Rename `comp-propagate'
-> `comp-fwprop'.
(comp-fwprop-prologue): Rename from `comp-propagate-prologue'.
(comp-fwprop-insn): Rename from `comp-fwprop-insn'.
(comp-propagate*): Rename from `comp-propagate*' and update.
(comp-fwprop): Rename from `comp-propagate' and update.
As explained previously, we can't simply make
eldoc-documentation-function an variable alias for
eldoc-documentation-strategy, because ElDoc is pre-loaded in Emacs <
28, where it holds at least one buffer-local binding. So if eldoc.el
is loaded in those versions, we do the variable alias binding in
reverse. We do this using a macro
eldoc--documentation-strategy-defcustom to at load time in which
direction to make the variable alias.
* lisp/emacs-lisp/eldoc.el
(eldoc--documentation-strategy-defcustom): Helper macro.
(eldoc-documentation-strategy, eldoc-documentation-function): Use it.
(Version): Bump to 1.5.0
When defining the obsolete variable alias for old
eldoc-documentation-function (which now points to the newer
eldoc-documentation-strategy), one gets the error "don't know how to
make a localized vareiable an alias". I'm not sure, but I suspect
this is because Eldoc is preloaded in Emacs 26.3 and the
eldoc-documentation-function variable is already set locally by some
Elisp buffer.
Uninterning the symbol shortly before defining the alias seems to fix
it.
* lisp/emacs-lisp/eldoc.el (eldoc-documentation-function):
Unintern on load.
(Version): Bump to 1.3.0
The command should always invoke Eldoc when called interactively,
instead of going through the usual checks, which are performed to
avoid interference with other commands.
* lisp/emacs-lisp/eldoc.el (eldoc-print-current-symbol-info): Rework.
(Version): Bump to 1.2.0
The truncation algorithm still has a long way to go for very narrow
frame sizes. It should become a generic mechanism that would allows
one to truncate a string so that fits in N possibly truncated screen
lines of a full-width window.
* lisp/emacs-lisp/eldoc.el (eldoc-handle-docs): Tweak
Add a simple pass to infer pure functions not explicitly declared as
such. Use this information only during compilation (speed 3) to
optimize out function calls whe possible.
