* lisp/dired-aux.el (dired-mark-read-string): Use read-from-minibuffer
instead of read-string when op-symbol is 'touch', since it's more clear
when read-from-minibuffer returns an empty string for RET,
and it was used in dired-mark-read-string initially (bug#70725).
* lisp/net/tramp-container.el: Move connection-local settings
where they belong to.
* lisp/net/tramp-sshfs.el (tramp-connection-properties): Do not
set connection property "direct-async-process".
(tramp-direct-async-process): Set connection-local value.
* test/lisp/net/tramp-tests.el (comp-warn-primitives): Declare.
(tramp-test18-file-attributes): Use it.
* lisp/calendar/todo-mode.el (todo-mode-map): Define it without a
value before its use in 'todo-insert-item--next-param' to pacify
the byte-compiler.
(todo-insert-item--next-param): Move back to its previous location
to keep it under the outline heading with related code.
* lisp/calendar/todo-mode.el (todo-insert-item--next-param): Move
to after the definition of 'todo-mode-map', which this function
uses. This avoids byte-compiler warning.
* lisp/calendar/todo-mode.el (todo-insert-item--next-param):
Initially assign each key in todo-mode-map a function identifying
it as invalid for item insertion, thus preventing mistakenly
pressing a key from executing an unwanted different todo-mode
command; the actual item insertion keys are redefined when looping
over the item insertion parameters.
* lisp/vc/vc.el: Add autoloaded setting of 'safe-local-variable' property
for 'vc-default-patch-addressee' exactly like it's already done
for 'vc-prepare-patches-separately'.
tab-line-mode should not inform the user of an unexpected change
when enabling the mode if already enabled. For example, when
running (tab-line-mode 1) repeatedly (bug#68765).
* lisp/tab-line.el (tab-line-mode): Modify case when user is informed.
Now, call all the custom providers for each step, using the provider
that moves point the smallest non-zero amount. This allows multiple
providers for a given "thing" to work nicely together.
* lisp/thingatpt.el (forward-thing-provider-alist): Update docstring.
(forward-thing): New implementation to call each provider N times.
(forward-thing-for-text-property): Take BACKWARD instead of N. Update
callers.
* test/lisp/thingatpt-tests.el (thing-at-point-providers)
(forward-thing-providers): Add more checks.
(consecutive-things-at-point): New test.
* lisp/tar-mode.el (tar-archive-from-tar): New local variable.
(tar-extract): Set it non-nil for an extracted member that happens
to be arc-mode archive.
* lisp/arc-mode.el (tar-archive-from-tar): Defvar it.
(archive-unique-fname): Make sure FNAME can be created in DIR,
even if FNAME is provided as an absolute file name (this happens
if the archive is a member of a Tar archive, for example).
(archive-extract): Set 'archive-remote' for archives that were
extracted from Tar archives. (Bug#70987)
* test/lisp/tar-mode-tests.el (tar-mode-test-tar-extract-zip-and-gz):
* test/lisp/arc-mode-tests.el (arc-mode-test-zip-extract-tar-and-gz):
New tests.
* test/data/decompress/tzg.tar.gz:
* test/data/decompress/ztg.zip:
New data files to test decompressing of different combinations
of nested tar and zip archives.
Previously, this always said "Previous" for
'previous-matching-history-element' (likewise "Next"). Now, the prompt
accounts for a negative prefix argument changing the search direction
(bug#70882).
* lisp/simple.el (previous-matching-history-element)
(next-matching-history-element): Consult numeric prefix argument to
determine the prompt string.
This also fixes an issue in EWW and bug-reference-mode where
(thing-at-point 'url) at the end of a URL would return nil.
See <https://lists.gnu.org/archive/html/emacs-devel/2024-02/msg00200.html>.
* lisp/thingatpt.el (forward-thing-provider-alist)
(bounds-of-thing-at-point-provider-alist): New variables...
(forward-thing, bounds-of-thing-at-point): ... use them.
(text-property-search-forward, text-property-search-backward)
(prop-match-beginning, prop-match-end): Declare.
(thing-at-point-for-text-property, forward-thing-for-text-property)
(bounds-of-thing-at-point-for-text-property): New functions.
* lisp/net/eww.el (eww--url-at-point): Use
'thing-at-point-for-text-property'.
(eww--bounds-of-url-at-point, eww--forward-url): New functions...
(eww-mode): ... use them.
* lisp/progmodes/bug-reference.el (bug-reference--url-at-point): Use
'thing-at-point-for-text-property'.
(bug-reference--bounds-of-url-at-point, bug-reference--forward-url): New
functions...
(bug-reference--init): ... use them.
* test/lisp/thingatpt-tests.el (thing-at-point-providers)
(forward-thing-providers, bounds-of-thing-at-point-providers): New
tests.
* etc/NEWS: Announce this change.
* src/regex-emacs.c (STORE_NUMBER): Make the arg int, not int16_t.
There’s no need for the caller to convert to int16_t, and using
int makes the machine code a bit smaller (and presumably a bit
faster) on x86-64 with GCC 14.
* src/lisp.h (hash_hash_t, knuth_hash): Use unsigned int and
unsigned long long int rather than uint32_t and uint64_t, as POSIX
does not guarantee the presence of uint64_t, and uint32_t and
uint64_t both in theory have problems with undefined behavior on
integer overflow. This doesn’t affect behavior (or even machine
code) on typical platforms.
* src/sfnt.c (sfnt_map_table, sfnt_read_table)
(sfnt_read_cvar_table): Pacify GCC -Wanalyzer-null-dereference.
The change to sfnt_read_cvar_table fixes what appears to be
an actual null-dereference bug.
When PIXELWISE is nil, we still calculate width in pixels and
then convert it to columns. However, part of the calculation
was using columns where it should have used pixels. (Bug#70894)
* config.bat: Generate stdbit.in-h from stdbit.in.h.
* msdos/sedlibmk.inp (GL_STDC_...): Enable generation
of ISO C2x features.
(STDBIT_H): Define to stdbit.h.
(GL_GENERATE_STDBIT_H_CONDITION): Define to 1.
* etc/ERC-NEWS: Mention new face `erc-information'.
* lisp/erc/erc-button.el (erc-button-add-buttons): Skip buttonization
when the "msg prop" `erc--skip' is present and contains the symbol
`button'. Set `alist' to nil in the same guard condition as a
roundabout way of suppressing further processing.
* lisp/erc/erc-networks.el (erc--insert-admin-message): Forward
declaration.
(erc-networks--insert-transplanted-content)
(erc-networks--transplant-buffer-content): Replace former with latter.
Change signature to take source and destination buffers as parameters.
(erc-networks--transplant-target-buffer-function): New function-valued
variable.
(erc-networks--target-transplant-in-progress-p): New variable, a flag
for downstream code to detect when a transplant is underway.
(erc-networks--reclaim-orphaned-target-buffers): Defer to
`erc-networks--transplant-target-buffer-function' to handle the actual
transplant business. Crucially, kill the buffer afterwards instead of
beforehand. If new buffer-association bugs emerge related to the
combining of old or renamed target buffers, this reordering may be at
fault.
(erc-networks--copy-over-server-buffer-contents): Pass old and new
buffers to `erc-networks--insert-transplanted-content'.
* lisp/erc/erc-stamp.el
(erc-stamp--defer-date-insertion-on-post-modify): Set `fn' slot of
`erc-stamp--date' instance to `ignore' when running the actual
callback in order to conserve a little space.
(erc-stamp--date-mode): Add and remove hook members for
`erc-networks--copy-server-buffer-functions' and
`erc-networks--transplant-target-buffer-function'.
(erc-insert-timestamp-left-and-right): Always clear
`erc-timestamp-last-inserted-right' to ensure a right stamp
accompanies every date stamp.
(erc-stamp--dedupe-date-stamps)
(erc-stamp--dedupe-date-stamps-from-buffer)
(erc-stamp--dedupe-date-stamps-from-target-buffer): New functions.
Date stamp behavior was revamped as part of bug#60936.
* lisp/erc/erc.el (erc-informational): New face.
(erc--insert-admin-message): New function to hide some "msg prop"
complexity from "upstream" libraries, like `erc-networks', and thus
avoid more forward-declarations. A less smelly approach would be to
devise a general interface owned by such libraries, or erc-common,
that `erc-mode' could then hook into on init.
(erc-display-message-highlight): Make face matching more limber to
accommodate the convention of face names lacking a "-face" suffix.
(erc-message-english-graft): New variable.
(erc-kill-channel): Inhibit execution of hook when
`erc-networks--target-transplant-in-progress-p' is non-nil.
* test/lisp/erc/erc-networks-tests.el
(erc-networks--rename-server-buffer--no-existing--orphan)
(erc-networks--rename-server-buffer--existing--reuse)
(erc-networks--rename-server-buffer--local-match)
(erc-networks--rename-server-buffer--local-nomatch): Use helper to
initialize markers.
* test/lisp/erc/erc-stamp-tests.el (erc-stamp--dedupe-date-stamps):
New test. (Bug#70928)
* etc/ERC-NEWS: Mention new flag `erc-killing-buffer-on-part-p' and
the renaming of `erc-kill-channel'.
* lisp/erc/erc-backend.el (erc-server-PART): Only kill a buffer on
behalf of `erc-kill-buffer-on-part' when the buffer hasn't already
been killed, and bind `erc-killing-buffer-on-part-p' to t when doing
so.
* lisp/erc/erc-log.el (erc-conditional-save-buffer): Don't save logs
when the buffer parameter is nil because that causes the server buffer
to be saved out. It's possible that user code relying on this
longstanding bug will be affected, however, by default, the server
buffer will also be saved out independently at designated junctures.
* lisp/erc/erc.el (erc-part-hook): Redo doc string.
(erc-killing-buffer-on-part-p): New variable, a flag to prevent
redundant execution of `erc-kill-channel-hook' members concerned with
parted channels.
(erc-kill-buffer-on-part): Tweak doc string.
(erc-kill-channel-hook): Use new name for `erc-kill-channel',
`erc-part-channel-on-kill'.
(erc-kill-channel, erc-part-channel-on-kill): Rename former to latter,
and inhibit execution when `erc-killing-buffer-on-part-p' is non-nil.
* test/lisp/erc/erc-scenarios-base-kill-on-part.el: New file.
(Bug#70840)
