* java/org/gnu/emacs/EmacsInputConnection.java
(EmacsInputConnection, performContextMenuAction): New function.
* java/org/gnu/emacs/EmacsNative.java (EmacsNative)
(performContextMenuAction): New function.
* src/android.c (android_get_gc_values): Implement more
efficiently.
* src/androidterm.c (android_handle_ime_event): Pass through
`update' argument to `finish_composing_text'. Fix thinko.
* src/textconv.c (really_finish_composing_text)
(really_set_composing_text, really_set_composing_region)
(handle_pending_conversion_events_1, finish_composing_text): New
argument `update'. Notify IME of conversion region changes if
set.
* src/textconv.h: Update structs and prototypes.
* src/androidterm.c (struct android_get_extracted_text_context):
New field `start_offset' and `end_offset'. Delete `offset'.
(android_get_extracted_text, android_build_extracted_text):
Replace `offset' with new args `start_offset' and `end_offset'.
(NATIVE_NAME): Set `start_offset' and `end_offset'.
(android_update_selection): Likewise.
* src/textconv.c (get_extracted_text): Likewise.
* src/textconv.h: Update prototypes.
It may not make much of a difference since the code path is probably
only taken once for i=0 with no more glyphs in the input, but this
change avoids depending on such external invariants and makes
the code less mystifying to the reader (bug#63807).
* lisp/emacs-lisp/bytecomp.el (important-return-value-fns):
Remove `delq` and `delete` since they resulted in false
positives that nobody was interested in silencing (bug#61730).
* test/infra/Dockerfile.emba (emacs-tree-sitter): Install only
grammars which are needed.
* test/infra/Makefile.in (subdir_template): Special handling of
progmodes.
(TREE-SITTER-FILES): New variable.
(tree-sitter-files-template): New rule.
(.PHONY, generate-test-jobs): Add it to dependencies.
* test/infra/gitlab-ci.yml (.tree-sitter-template):
Add test/lisp/progmodes/*-ts-mode-resources/**.
(test-tree-sitter): Extends also .tree-sitter-files-template. Use
$tree-sitter-files.
* test/infra/test-jobs.yml: Regenerate.
* java/debug.sh (is_root): Go back to using unix sockets; allow
adb to forward them correctly.
* java/org/gnu/emacs/EmacsInputConnection.java
(getExtractedText): Don't print text if NULL.
* java/org/gnu/emacs/EmacsService.java (EmacsService): New field
`imSyncInProgress'.
(updateIC): If an IM sync might be in progress, avoid deadlocks.
* java/org/gnu/emacs/EmacsView.java (onCreateInputConnection):
Set `imSyncInProgress' across synchronization point.
* src/android.c (android_check_query): Use __atomic_store_n.
(android_answer_query): New function.
(android_begin_query): Set `android_servicing_query' to 2.
Check once, and don't spin waiting for query to complete.
(android_end_query): Use __atomic_store_n.
(android_run_in_emacs_thread): Compare-and-exchange flag. If
originally 1, fail.
* src/textconv.c (really_set_composing_text): Clear conversion
region if text is empty.
The call to `format-message` had no effect in itself, but it turns out
that actually emitting the result just results in useless warnings so
we remove it entirely.
* lisp/xml.el (xml-parse-attlist): Don't attempt to collapse multiple
consecutive whitespace characters into a single space, which is wrong
to do a this point when attributes must be assumed to be CDATA.
The code was actually unintentionally correct since it forgot to use
the return value of `replace-regexp-in-string` (bug#63740).
Despite its name, `any` does not match any character (use `anychar`
for that) but is an old synonym for `not-newline` and `nonl`,
retained for compatibility with the obsolete `sregex` package.
(In SRE as defined by SRFI-115, `any` does match any character.)
There is also a mild but unnecessary confusion with the more
frequently used `any` construct for character alternatives.
* lisp/edmacro.el (edmacro-mode-font-lock-keywords):
* lisp/erc/erc.el (erc--ensure-url):
* lisp/org/org.el (org-fontify-meta-lines-and-blocks-1):
* lisp/progmodes/scheme.el (scheme-imenu-generic-expression):
* test/lisp/find-cmd-tests.el (find-cmd-test-find-cmd):
Replace uses of `any` with `nonl`.
* lisp/emacs-lisp/bytecomp.el
(byte-compile--suspicious-defcustom-choice): Rename to...
(byte-compile--defcustom-type-quoted): ...this and rewrite to make
more sense. All callers updated.
(byte-compile-nogroup-warn): Better warning message.
* test/lisp/emacs-lisp/bytecomp-tests.el
(test-bytecomp-defgroup-choice): This never failed because it wasn't
actually a test. Turn it into...
(bytecomp-test-defcustom-type-quoted): ...this, which is.
* doc/misc/modus-themes.org: Update the manual.
* etc/themes/modus-operandi-tritanopia-theme.el:
* etc/themes/modus-vivendi-tritanopia-theme.el: Add new themes
optimized for users with blue-yellow color deficiency (tritanopia).
* etc/themes/modus-operandi-deuteranopia-theme.el:
* etc/themes/modus-operandi-theme.el:
* etc/themes/modus-operandi-tinted-theme.el:
* etc/themes/modus-vivendi-deuteranopia-theme.el:
* etc/themes/modus-vivendi-theme.el:
* etc/themes/modus-vivendi-tinted-theme.el: Update theme files.
* etc/themes/modus-themes.el: Update main file to the latest version.
* etc/NEWS: Rephrase the announcement about the new Modus themes to
also mention the tritanopia-optimized items.
Detailed release notes here:
<https://protesilaos.com/codelog/2023-05-30-modus-themes-4-2-0/>.
* etc/ERC-NEWS: Mention behavioral changes to functionality provided
by the `truncate' and `log' modules and also the "/CLEAR" command.
* lisp/erc/erc-log.el (erc-truncate-buffer-on-save): Deprecate option
because three's a crowd, and ERC already has a dedicated module as
well as a slash command for this purpose. And although this commit
restores functionality, this option has been unusable since at least
the release of ERC 5.5, with no known complaints received thus far.
Also, the doc string of `erc-save-buffer-in-logs' makes no mention of
this feature WRT interactive invocations or otherwise.
(erc-log-mode, erc-log-enable, erc-log-disable): Subscribe to
new internal hook `erc--pre-clear-functions'.
(erc-log--save-in-progress-p): New variable to help restore
`erc-truncate-buffer-on-save' and promote code reuse.
(erc-logging-enabled): Guard with `erc-log--saved-in-progress-p'.
(erc-save-buffer-in-logs): Overload `buffer' parameter to allow
various hooks to supply a non-buffer as well. Warn when people use
`erc-truncate-buffer-on-save', which is now deprecated.
* lisp/erc/erc-stamp.el (erc-stamp-mode, erc-stamp-enable,
erc-stamp-disable): Subscribe to `erc--pre-clear-functions'.
(erc-stamp--update-saved-position): New function for updating
last-logged marker on `erc-stamp--insert-date-function'.
(erc-stamp--reset-on-clear): New function to forget last inserted
stamps when truncating.
* lisp/erc/erc-truncate.el (erc-truncate-mode, erc-truncate-enable,
erc-truncate-disable): Use `erc-insert-done-hook' instead of
`erc-insert-post-hook', as implicitly suggested by an ancient comment,
which ponders whether truncating the buffer at the insertion phase may
be harmful to other hook members.
(erc-truncate-buffer-to-size): Set truncation boundary at message
break instead of line break. Run `erc--pre-clear-functions'.
(erc-truncate-buffer): Save excursion. This should probably be
handled by `erc-truncate-buffer-to-size' instead, but that's
likelier to cause breakage in third-party code.
* lisp/erc/erc.el (erc--pre-clear-functions): New internal hook.
(erc-cmd-CLEAR): Run `erc--pre-clear-functions' before clearing, and
don't blow away prompt. The latter was a regression caused by
05f6fdb9e7 "Preserve ERC prompt and its bounding markers".
* test/lisp/erc/erc-scenarios-log.el: New file. (Bug#60936)
* lisp/erc/erc.el (erc--refresh-prompt): New function for redrawing
the prompt in a couple select places.
(erc-display-line-1, erc-display-msg): Replace the prompt after
inserting messages.
* test/lisp/erc/erc-tests.el (erc--refresh-prompt): New
test. (Bug#60936)
* etc/ERC-NEWS: Improve section mentioning `erc-tab'.
* lisp/erc/erc-button.el (erc-button-alist): Restore a commented out
version of the "<URL ...>" entry, which was dropped in 5adda2f468
"Revise FORM-as-function interface in erc-button-alist" for being
redundant. This may help unlikely objectors better adapt to the
churn.
(erc-button-next, erc-button-previous): Don't bother accommodating raw
prefix arguments; expect numeric arguments only.
* lisp/erc/erc.el (erc-tab): Change interactive spec to lowercase
"p". (Bug#60933)
* doc/misc/erc.texi: Add new SOCKS section to the chapter on
Integrations. Also bump version in front matter.
* lisp/erc/erc.el (erc-open-socks-tls-stream): New convenience
function to connect over Tor via SOCKS proxy.
* lisp/emacs-lisp/comp-cstr.el (comp--compute-typeof-types)
(comp--compute--pred-type-h): New functions.
(comp-cstr-ctxt): Make use of.
(comp-cstr-ctxt-update-type-slots): New function.
* lisp/emacs-lisp/comp.el (comp-spill-lap): Use
`comp-cstr-ctxt-update-type-slots'.
* java/org/gnu/emacs/EmacsNative.java (EmacsNative): New
function `blitRect'.
* java/org/gnu/emacs/EmacsSurfaceView.java (EmacsSurfaceView):
Use it on Android 8.x.
* src/android.c (blitRect): Implement new function.
Fix unstable proced test by omitting the revert parameter in
'proced-update'. This was caused by the process being refined on
exiting between the initial 'proced' call and the successive
'proced-update' call. This resulted in proced skipping the refinement
in 'proced-update', causing all processes to be shown again and the
test to fail.
* test/lisp/proced-tests.el (proced-refine-with-update-test): Do not
use revert parameter when calling 'proced-update'.
* lisp/net/tramp-sh.el (tramp-sh-handle-process-file): Set `env'
explicitly.
(tramp-barf-if-no-shell-prompt): Still add optional trailing
`ansi-color-control-seq-regexp'.
* lisp/net/tramp.el (tramp-make-tramp-file-name): Keep hop while
in file name completion.
* test/lisp/net/tramp-tests.el
(tramp-test26-interactive-file-name-completion): Adapt test.
