The basic idea is, we're moving the setting/unsetting of
`gnus-registry-enabled' so that it is a more reliable indicator of
whether the registry is actually available or not.
* lisp/gnus/gnus-registry.el (gnus-registry-load): Move the setting of
`gnus-registry-enabled' here, after we're sure the registry actually
has been loaded.
(gnus-registry-initialize): Move it out of here. All this function
does is conditionally call `gnus-registry-load'.
(gnus-registry-register-message-ids): Only check
`gnus-registry-enabled', since we'd more confident about this
now. Sort checks in order of increasing complexity.
(gnus-registry-clear): Unset `gnus-registry-enabled'.
(gnus-registry-install-hooks): This function should only install
hooks, not set variables.
(gnus-registry-unload-hook): Change the unload hook to call
`gnus-registry-clear', as that will do all the necessary
work (including calling `gnus-registry-unload-hook')
* lisp/isearch.el (isearch-emoji-by-name): Add a new command to insert
Emoji characters into incremental search strings. (Bug#52605)
(isearch-mode-map): Bind it to 'C-x 8 e RET'.
(isearch-menu-bar-map): Add it to the menu bar.
* doc/emacs/search.texi (Special Isearch): Update the documentation to
mention the new command.
* etc/NEWS: And advertise it.
Lambda-lifted variables (ones passed explicitly to lambda-lifted
functions) that are also captured in an outer closure and shadowed
were renamed incorrectly (bug#51982).
Reported by Paul Pogonyshev.
* lisp/emacs-lisp/cconv.el (cconv--lifted-arg): New.
(cconv-convert): Provide correct definiens for the closed-over
variable.
* test/lisp/emacs-lisp/bytecomp-tests.el (bytecomp-tests--test-cases):
* test/lisp/emacs-lisp/cconv-tests.el (cconv-tests--intern-all)
(cconv-closure-convert-remap-var): Add tests.
(cherry picked from commit 45252ad8f9)
* lisp/eshell/esh-opt.el (eshell--split-switch): New function.
(eshell-set-option): Allow setting a supplied value instead of always
consuming from 'eshell--args'.
(eshell--process-option): Support consuming option values specified as
a single token.
(eshell--process-args): For short options, pass full switch token to
'eshell--process-option'.
* test/lisp/eshell/esh-opt-tests.el (esh-opt-process-args-test): Fix
test.
(test-eshell-eval-using-options): Add tests for various types of
options.
* doc/misc/eshell.texi (Defining new built-in commands): New
subsection, describe how to use 'eshell-eval-using-options'.
* etc/NEWS: Announce the change.
* src/font.c (font_score, font_delete_unmatched): Don't assume
weight, slant, and width properties of the font must be fixnums:
some variable-weight fonts violate that assumption. Reported
by Sean Whitton <spwhitton@spwhitton.name>. This is for builds
with Fontconfig < v2.11.91, for which the fix in
'ftfont_pattern_entity' cannot work. (Bug#52888)
* test/lisp/cedet/semantic/bovine/gcc-tests.el
(semantic-gcc-test-output-parser-this-machine): Also detect Apple
clang on macOS Monterey.
(cherry picked from commit 6e52becfbe)
* test/lisp/progmodes/flymake-tests.el (flymake-tests--gcc-is-clang)
(different-diagnostic-types, included-c-header-files): Skip tests that
depend on the `gcc` command really being GCC and not Clang.
(cherry picked from commit b2167d9843)
* lisp/gnus/gnus-art.el (gnus-treat-emphasize): Don't do emphasis
in text/html parts, because this leads to mis-filled text
(especially with *...* that's boldified, which typically becomes
much wider).
This is considered a temporary fix until the cause of
fringe-overwriting underlines is established.
* src/xterm.c (x_draw_glyph_string): Don't constrain decorations
if the row is a tab or modeline.
* lisp/window.el (display-comint-buffer-action):
Use display-buffer--same-window-action as the default value.
* lisp/progmodes/sh-script.el (sh-show-shell):
* lisp/shell.el (shell): Add display-comint-buffer-action to pop-to-buffer.
* lisp/progmodes/project.el (project-shell, project-eshell): Use
`bound-and-true-p display-comint-buffer-action' for backward-compatibility.
https://lists.gnu.org/archive/html/emacs-devel/2021-12/msg02964.html
Reduce dependency on Haiku internals, by not assuming that
Haiku errno values (which are negative) are neither -1 nor -2.
This removes an #ifdef HAIKU while still maintaining
portability to Haiku.
* src/filelock.c (NEGATIVE_ERRNO, ANOTHER_OWNS_IT, I_OWN_IT):
New constants, which should work regardless of whether
we are on Haiku or B_USE_POSITIVE_POSIX_ERRORS is defined.
(current_lock_owner, lock_if_free, lock_file, unlock_file)
(Ffile_locked_p): Use them, without assuming anything about errno
value sign.
This breaks when x_free_xi_devices is called from a
HierarchyChange event where a master device has been removed or
a slave device has been detached, and it should no longer be
required since we now rely on passive grabs.
* src/xterm.c (x_init_master_valuators): Stop calling
XIUngrabDevice on all devices.
When calling outline-on-heading-p, set its argument INVISIBLE-OK to t,
so when <backtab> (bound to outline-cycle-buffer) hides the current outline
at a deep level, <backtab> should continue cycling outlines even when
the current outline heading is invisible. It worked this way before
the recent change that moved cycling keys from font-lock text properties
to outline-minor-mode keymap with :filter.
* test/lisp/files-tests.el (files-tests--with-buffer-offer-save):
Turn it into a function.
(files-tests-save-buffers-kill-emacs--asks-to-save-buffers)
(files-tests-buffer-offer-save): Adjust calls accordingly
* test/lisp/auth-source-tests.el (auth-source-test-searches):
(auth-source-test-secrets-create-secret):
(auth-source-test-netrc-create-secret, auth-source-delete):
Use 'auth-info-password' instead of ad hoc code.
* doc/misc/auth.texi (Help for developers): Document
'auth-source-pick-first-password' and 'auth-info-password'.
Update the 'auth-source-search' example.
