* lisp/progmodes/project.el (project-find-file): Don't say
anything about recognizing file names, as that may lead the user
to think that it'll check whether the string at point is an
existing file (bug#44588).
* src/keymap.c (Fdescribe_buffer_bindings): Fix a call in
describe-buffer-bindings. This fixes a mistake in my previous commit
to prefer the Lisp version of describe-map-tree (8a1441310a), where
0 was accidentally converted to Qt in two places.
* lisp/dired-aux.el (dired-query): Replace read-char-choice call
with read-char-from-minibuffer.
* lisp/subr.el (read-char-choice): Restore the previous version
that uses read-key.
(read-char-from-minibuffer): Bind help-char to help-form-show
when help-form is non-nil.
* lisp/progmodes/python.el
(python-font-lock-keywords-maximum-decoration): `symbol-name'
should not be quantified by a `+' as it is redundant and performs
very badly (bug#44572).
* lisp/cedet/semantic.el (semantic-submode-list):
(semantic-default-submodes): Remove mention of
global-cedet-m3-minor-mode, which no longer exists, apparently
(bug#44565).
* lisp/progmodes/python.el
(python-font-lock-keywords-maximum-decoration): Fix regular
expressions for font lock of assignments with type hints (bug#44568).
The font lock of assignments with type hints in Python is rather bad.
Consider the following example:
from typing import Mapping, Tuple, Sequence
var1: int = 5
var2: Mapping[int, int] = {10: 1024}
var3: Mapping[Tuple[int, int], int] = {(2, 5): 32}
var4: Sequence[Sequence[int]] = [[1], [1, 2], [1, 2, 3]]
var5: Sequence[Mapping[str, Sequence[str]]] = [
{
'red': ['scarlet', 'vermilion', 'ruby'],
'green': ['emerald green', 'aqua']
},
{
'sword': ['cutlass', 'rapier']
}
]
As things stand right now, only ‘var1’ would be highlighted. To make
things worse, the ‘Mapping’ type hint of ‘var2’ would also be
highlighted, which is entirely incorrect.
This commit makes all of ‘var1’ through ‘var5’ be highlighted
correctly.
* test/lisp/progmodes/cperl-mode-tests.el
(cperl--run-test-cases): New macro, factored out from various
indentation / rewriting tests. Contains documentation of the
format used by the cperl-mode-resources files.
(cperl-test-bug-19709): Replace 'next-line' by 'forward-line'.
(cperl-test-indent-exp),
(cperl-test-indent-styles),
(cperl-test-bug-30393): Use the new macro.
(cperl-test-bug-19709): Make fit for Emacs 26.
(cperl-test-indent-styles): Skip for Perl mode (bug#44561).
(server--create-frame): New function, extracted from
`server-create-dumb-terminal-frame`.
(server-create-window-system-frame, server-create-tty-frame):
(server-create-dumb-terminal-frame): Use it.
(server-create-dumb-terminal-frame): New function.
(server-process-filter): Use it.
(server-delete-client): Don't delete tty terminal when it's not
exclusive to this client.
* lisp/org/org-element.el (org-element-parse-secondary-string):
Don't set buffer-local variables buffer-file-name and buffer-file-truename
in temporary buffer.
* doc/emacs/basic.texi (Inserting Text):
Mention transient input method "iso-transl".
* lisp/leim/quail/latin-ltx.el: Use same Keywords as in other quail files.
Prefer vc-state to benefit from its caching (bug#44534)
This same technique is used in vc-git.el,
per commit 2018-06-28 "Remove extra process call from vc-git-find-file-hook"
93c41ce6aa.
* lisp/vc/vc-hg.el (vc-hg-registered): Use vc-state rather than vc-hg-state.
* lisp/gnus/nnselect.el (nnselect-request-group): If an ephemeral
group is empty, there is nothing to see, so remove the group.
(nnselect-run): Catch and return an empty artlist on error.
* lisp/emacs-lisp/shortdoc.el (sequence): use seq-contains-p instead
of seq-contains, which is obsolete as of 27.1. (Bug#44536)
Copyright-paperwork-exempt: yes
* test/lisp/emacs-lisp/rx-tests.el (rx-compat): Use with-no-warnings
instead of with-suppressed-warnings which complains when running
the test interactively.
This is a regression from Emacs 26.
Reported by Phillip Stephani.
* lisp/emacs-lisp/rx.el (rx--pcase-transform): Process ? and ?? correctly.
* test/lisp/emacs-lisp/rx-tests.el (rx-pcase): Add test case.
This is a followup to last change in image.c.
* src/image.c (rsvg_handle_get_intrinsic_dimensions): Define to
call fn_rsvg_handle_get_intrinsic_dimensions.
(svg_css_length_to_pixels): Compile only for librsvg >= 2.46.0, as
RsvgLength type was not defined before.
* src/nsterm.m (ns_destroy_window): Close the window before freeing
the frame resources so we don't end up accessing the frame struct
after it's been freed.
* lisp/progmodes/cperl-mode.el (cperl-style-alist): Add
cperl-close-paren-offset to the settings for PBP style.
* test/lisp/progmodes/cperl-mode-tests.el (cperl-bug19709):
New test to verify correct indentation of closing parentheses (Bug#19709).
* test/lisp/progmodes/cperl-mode-resources/cperl-bug-19709.pl:
New test case with code from the bug report.
* test/lisp/progmodes/cperl-mode-resources/cperl-indent-styles.pl:
Add a new test clause for cperl-close-paren-offset.
* test/src/keymap-tests.el
(keymap-keymap-set-parent/returns-parent): Remove test for the return
value of set-keymap-parent. It is not clear that returning the value
is a very good idea.
Problem pointed out by Stefan Monnier <monnier@iro.umontreal.ca>.
* lisp/gnus/gnus-group.el (gnus-group-make-search-group,
gnus-group-read-ephemeral-search-group): If the query is coming in via
the old 'nnir-query-spec key, we know not to parse it.
* lisp/gnus/gnus-search.el (gnus-search-make-query-string): Check if
the query was sent in as '(query "query"), and not '(query . "query).
(gnus-search-imap-search-keys): Add x-gm-raw to imap search keys.
(gnus-search-prepare-query): If we know this query should be raw,
don't even try parsing it, as it probably won't work.
* lisp/progmodes/project.el (project--write-project-list):
Let-bind print-length and print-level to nil to not truncate the
saved project list with unreadable ellipsis.
* lisp/gnus/gnus-search.el (gnus-search-server-to-engine): Because of
the way we've set up the obsolete variable alias for
`nnir-method-default-engines', we may end up with its value in
`gnus-search-default-engines'. Make the check for "old style" values
general, so we catch them no matter where they came from.
