Treat `ignore` as any other function during source-level optimisation,
to avoid having its warning-suppression effects cancelled by repeated
passes. Instead, define a custom code generation function.
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker):
Don't treat `ignore' specially here.
(side-effect-free-fns): Don't mark `ignore` as side-effect-free
or error-free (although it is), since that would allow the optimiser
to elide calls.
* lisp/emacs-lisp/bytecomp.el (ignore, byte-compile-ignore):
Define and register a code-gen function.
Optimise `member` and `assoc` (etc) with constant empty list
* lisp/emacs-lisp/byte-opt.el
(byte-optimize-assq): New.
(byte-optimize-member, byte-optimize-assoc, byte-optimize-memq):
When the list argument is constant nil, the result is always nil.
* test/lisp/emacs-lisp/bytecomp-tests.el (bytecomp-tests--test-cases):
Add test cases.
* lisp/dired-aux.el (dired-do-find-regexp-and-replace):
Disregard the customized value of xref-auto-jump-to-first-xref
(it breaks the xref-query-replace-in-results invocation).
* lisp/progmodes/xref.el (xref-auto-jump-to-first-xref):
Mention the caveat for users or xref distributed through ELPA.
Allen Li [Mon, 6 Sep 2021 08:59:54 +0000 (10:59 +0200)]
Make `gnus-group-use-permanent-levels' work better
* lisp/gnus/gnus-group.el (gnus-group-default-level): Removed implicit
setting of gnus-group-use-permanent-levels.
(gnus-group-list-groups): Set gnus-group-use-permanent-levels
explicitly.
(gnus-group-get-new-news): Set gnus-group-use-permanent-levels
explicitly (bug#50417).
(gnus-group-get-new-news): When the user has given a numeric
prefix, use that as the list level.
project--files-in-directory: Fix handling of ignores
* lisp/progmodes/project.el (project--files-in-directory):
Pass "." as the DIR argument to 'find' because otherwise the ignore
expression can match the project root directory name, which we don't
want to happen (bug#50240). Fixup the resulting file names at the end
with concatenation.
Originally I thought it could lead to worse performance, but the
results show equal or slightly better timings.
* lisp/progmodes/xref.el (xref-matches-in-directory):
Apply a similar fix.
(xref--find-ignores-arguments): Use file-name-as-directory, so
that when passed "." replace-match still had the expected effect.
* test/lisp/progmodes/project-tests.el (project-ignores-bug-50240):
New test.
* test/lisp/progmodes/xref-tests.el
(xref-matches-in-directory-filters-with-ignores): New test.
Improve the documentation around the read-key/minibuffer prompting
* lisp/subr.el (read-char-choice-use-read-key): Explain the
difference.
(read-char-choice): Mention the variables.
(y-or-n-p-use-read-key): Explain the difference.
(y-or-n-p): Mention the variable (bug#50390).
Eli Zaretskii [Sun, 5 Sep 2021 07:59:07 +0000 (10:59 +0300)]
Improve documentation of new Xref options
* doc/emacs/maintaining.texi (Looking Up Identifiers):
* etc/NEWS: Document the new Xref user options. Move a lost
Xref-related entry to the rest of them.
Augusto Stoffel [Sat, 4 Sep 2021 09:16:11 +0000 (11:16 +0200)]
Fixes for 'python-shell-send-string' and 'python-shell-send-file'
* lisp/progmodes/python.el (python-shell-send-string): use a temporary
file for sufficiently long strings.
(python-shell-send-file, python-shell-eval-file-setup-code): Avoid
showing "plumbing code" in the traceback (bug#32042).
Ian W [Sat, 4 Sep 2021 03:53:24 +0000 (20:53 -0700)]
Fix error handling in 'ispell-init-process'
* lisp/textmodes/ispell.el (ispell-init-process): When Ispell
initialization reports an error, call 'ispell-accept-output' only
if the Ispell process is still alive. (Bug#50370)
Michael Albinus [Sun, 5 Sep 2021 07:23:30 +0000 (09:23 +0200)]
Cleanup tramp-tests.el
* test/lisp/net/tramp-tests.el
(tramp--test--deftest-direct-async-process): Fix macro declaration.
(tramp-test29-start-file-process, tramp-test30-make-process):
Do not run connection type test for direct async processes.
Xref: automatic jumping to the first definition or reference
* lisp/progmodes/xref.el (xref-auto-jump-to-first-definition)
(xref-auto-jump-to-first-xref): New user options.
Discussed in bug#50067.
(xref--show-xrefs, xref--show-defs): Use them here.
(xref--auto-jump-first): New function, handles different values.
(xref-show-definitions-buffer)
(xref-show-definitions-buffer-at-bottom): Use it.
Eli Zaretskii [Sat, 4 Sep 2021 10:34:32 +0000 (13:34 +0300)]
Improve documentation of line truncation and wrapping
* doc/emacs/display.texi (Line Truncation, Visual Line Mode):
* doc/lispref/display.texi (Truncation): Document that turning on
line truncation disables wrapping, and vice versa.
* src/buffer.c (syms_of_buffer) <truncate-lines>:
* src/xdisp.c (syms_of_xdisp) <truncate-partial-width-windows>:
Warn against turning on when 'visual-line-mode' is in effect.
* lisp/simple.el (visual-line-mode): Document that this mode
disables line truncation. (Bug#29664)
Add support for customization group hyperlinks in doc strings
* lisp/help-mode.el (help-customization-group): New button.
(help-xref-customization-group-regexp): New const.
(help-make-xrefs): Use them to allow making customization group
buttons.
* lisp/hi-lock.el (hi-lock-mode): Tweak the doc string so that
running hi-lock-mode in a *Help* buffer showing this help text
won't issue an error (bug#20977).
* doc/emacs/calendar.texi: Document the change.
* lisp/calendar/diary-lib.el (diary-offset):
* test/lisp/calendar/icalendar-tests.el: Add a test (bug#50195).
Augusto Stoffel [Fri, 3 Sep 2021 12:26:51 +0000 (14:26 +0200)]
Change Python eval to send directly instead of using temporary files
* lisp/progmodes/python.el (python-shell-eval-setup-code): New
const for setting up eval (bug#49822).
(python-shell--encode-string): New function.
(python-shell-send-string): Use it to send commands directly
instead of writing to a temporary file.
(python-shell-send-string-no-output): Adjust sending.
(python-shell-send-file): Ditto.
* css-mode.el (css-property-alist, css-value-class-alist): Provide new
position related entries for use in css-property-alist. Make sure
that property-alist reflect the spec:
https://www.w3.org/TR/css-align-3/#property-index
Fix `describe-function' for autoloaded adviced functions
* lisp/emacs-lisp/nadvice.el (advice--make-single-doc): Factor
out.
(advice--make-docstring): From here (bug#23523). Also include
advices for autoloads.
Make recent timer changes more backwards-compatible
* lisp/emacs-lisp/timer.el (timerp, timer-event-handler): Make
backwards-compatible with old .elc files that have their own
versions of `timer-create'd structures.
Stefan Kangas [Wed, 28 Jul 2021 11:31:40 +0000 (13:31 +0200)]
Move cl-lib tests for caaXr to subr-tests.el
* test/lisp/emacs-lisp/cl-lib-tests.el (cl-test-caaar)
(cl-test-caadr): Move tests using a cl-lib alias from here...
* test/lisp/subr-tests.el (subr-test-caaar, subr-test-caadr): ...to
here. The functions under test are in subr.el.
* lisp/textmodes/css-mode.el (css-property-alist): Consolidate
`align-{contents, items, self}', as well as the corresponding
values for `justify-{contents, items, self}' and `place-{contents,
items, self}'. Values extracted from the flex part and into its
own block (bug#50345).
Formerly, bug-reference-setup-from-vc-alist basically had one entry
per host (like gitlab.com). Restructure so that it's easy to add new
hosts being just an instance of some type of forge such as SourceHut,
Gitea, or GitLab.
While we're at it, add support for gitea.com, salsa.debian.org, and
framagit.org, the latter two being GitLab instances.
* lisp/progmodes/bug-reference.el (bug-reference-gitea-instances)
(bug-reference-gitlab-instances,bug-reference-sourcehut-instances):
New variables listing online instances of those forges.
(bug-reference--setup-from-vc-alist): New function (and variable for
caching) using the former three new variables to generate suitable VC
auto-setup alist.
(bug-reference-try-setup-from-vc): Use both
bug-reference-setup-from-vc-alist and
bug-reference--setup-from-vc-alist.
Eli Zaretskii [Thu, 2 Sep 2021 11:58:41 +0000 (14:58 +0300)]
Fix 'clone-frame' on TTY frames
* src/frame.c (Fmake_terminal_frame): Make a separate copy of the
faces for the new frame before calling modify-frame-parameters, as
on TTY frames that needs the faces already set up. (Bug#34715)
Eli Zaretskii [Thu, 2 Sep 2021 10:43:00 +0000 (13:43 +0300)]
Improve a recent change in 'mule-cmds.el'
* lisp/international/mule-cmds.el
(select-safe-coding-system--format-list): Display codepoints in
hex and as characters. Display the ellipsis in a better place.
Enlarge the limit to 5 codepoints, as the display is now easier to
grasp.
(select-safe-coding-system-interactively): Don't use
'select-safe-coding-system--format-list' for "rejected"
coding-systems, as those don't come with characters and positions.
* lisp/international/mule-cmds.el
(select-safe-coding-system--format-list): New function to display
the coding system information in a more readable format (bug#31062).
(select-safe-coding-system-interactively): Use it.
* lisp/emacs-lisp/lisp.el (backward-up-list): Reformat the doc
string into several paragraphs (bug#31349).
(up-list): Ditto and fix the "start" to be "end".
Eli Zaretskii [Wed, 1 Sep 2021 16:27:43 +0000 (19:27 +0300)]
Fix segfault with invalid key-translation-map binding
* src/keyboard.c (access_keymap_keyremap): Don't assume an invalid
function is specified as a symbol. Reported by Perry E. Metzger
<perry@piermont.com>.