* src/keyboard.c (kbd_buffer_get_event): Ignore selection-request
events when accessing the frame of an event. Suggested by dick
r. chiang <dick.r.chiang@gmail.com>. (Bug#80558)
Support calls to ns_lisp_to_color or Fns_list_colors in headless
environments where ns_term_init is not called, and calls before
ns_term_init runs.
* src/emacs.c (main): Call ns_init_colors when not in dump mode.
* src/nsfns.m (ns_init_colors): Make public function.
(Fns_list_colors): Remove ns_init_colors call.
* src/nsterm.h: extern ns_init_colors.
* src/frame.c (delete_frame): Call 'get-mru-frame' safely and
with EXCLUDE-FRAME argument set up correctly.
* lisp/frame.el (get-mru-frames, get-mru-frame): Fix doc-strings.
* src/json.c (json_out_string):
Optimise for runs of non-special ASCII chars.
This seems to be a a substantial performance gain for long strings but a
smaller regression for short strings (up to 6 chars or thereabouts,
depending on cpu, compiler, and/or C library). Still likely worth it.
Suggested by Pavel <cyberkm@gmail.com>.
See initial discussion here:
https://lists.gnu.org/archive/html/emacs-devel/2026-02/msg00400.html
* lisp/international/characters.el (pairs): Populate
special-mirror-table.
* src/xdisp.c (produce_special_glyphs): Add two arguments. One
to identify the paragraph direction and one to identify that the
glyph is produced on the left hand side of a window. Mirror
glyph defined in the display table according to the new
special-mirror-table. Bidi mirroring always takes precedence.
(init_iterator, insert_left_trunc_glyphs, display_line)
(display_string): Call 'produce_special_glyphs' with new arguments.
(syms_of_xdisp) <special-mirror-table>: New char-table.
* etc/NEWS: Announce the change.
* src/w32image.c (w32_gdip_export_frame): New function.
(gdiplus_init): Load 'GdipCreateBitmapFromHBITMAP' from GDI+ DLL.
* src/w32fns.c (Fw32_export_frame): New primitive.
* etc/NEWS: Announce the new primitive.
* src/keyboard.c (syms_of_keyboard):
<multiple_terminals_merge_keyboards>: New variable.
(read_event_from_main_queue): Mention it in message.
* lisp/cus-start.el: Make it customizable.
* src/callint.c (Ffuncall_interactively):
* src/keyboard.c (Frecursive_edit):
* src/minibuf.c (read_minibuf): When its value is true, don't
enter single-kboard mode (bug#79892).
* admin/notes/multi-tty:
* etc/NEWS: Document it.
Also, document stuffing better.
* src/keyboard.c (stuff_buffered_input):
Give up on stuffing if it fails.
* src/sysdep.c (stuff_char): Return failure indication.
* src/keyboard.c (kbd_buffer_get_event): Pass the event's frame
up to the caller by means of a new 'struct frame **' argument.
(read_event_from_main_queue): Show a message in locked frames in
single-kboard mode (bug#79892).
* src/xdisp.c (log_message): Factor out of message3.
(message3): Call it.
(message3_nolog): Rename to ...
(message3_frame_nolog): ... this. New 'struct frame *' argument
which causes temporarily switching to another frame when
displaying the message.
(message3_frame, message3_nolog): New functions.
* src/lisp.h: Declare message3_frame and message3_frame_nolog.
* admin/notes/multi-tty: Remove notes on showing a message.
* lisp/subr.el (command-line-max-length): Move from here ...
* src/callproc.c (syms_of_callproc): <Vcommand_line_max_length>:
... to here. Initialize by calling sysconf(3) if possible.
macOS 26 introduced new event processing behavior that causes scrolling
lag and input handling problems in Emacs. This patch disables two
features via NSUserDefaults when built against the macOS 26 SDK:
- NSEventConcurrentProcessingEnabled
- NSApplicationUpdateCycleEnabled
This fix is based on the equivalent patch in emacs-mac by Mitsuharu
Yamamoto. See: https://bitbucket.org/mituharu/emacs-mac/commits/e52ebfd
* src/nsterm.m (ns_term_init): Disable problematic event processing
when built for macOS 26+.
* src/lisp.h (FOR_EACH_TAIL_INTERNAL): Divvy up the code into...
(FOR_EACH_TAIL_BASIC, FOR_EACH_TAIL_STEP_CYCLEP): ...these macros,
so that they can be used in more flexible ways.
* src/fns.c (internal_equal_1): Detect circular lists and call...
(internal_equal_cycle): ...this function that keeps comparing
but now detecting cycles in the other argument.
* lisp/emacs-lisp/testcover.el (testcover-after):
Remove unnecessary error handling.
* test/src/fns-tests.el (test-cycle-equal): Adapt and extend.
* test/lisp/emacs-lisp/testcover-resources/testcases.el
(testcover-testcase-cyc1): Remove case that no longer applies.
* doc/lispref/objects.texi (Equality Predicates): Update.
* etc/NEWS: Announce.
* admin/merge-gnulib (GNULIB_MODULES): Add crypto/sha3-buffer.
* lib/sha3.c: New file, imported by running admin/merge-gnulib.
* lib/sha3.h: Likewise.
* m4/sha3.m4: Likewise.
* lib/gnulib.mk.in: Updated by admin/merge-gnulib.
* m4/gnulib-comp.m4: Likewise.
* src/fns.c: Include sha3.h
(Fsecure_hash_algorithms): Add Qsha3_224, Qsha3_256, Qsha3_384, and
Qsha3_512.
(secure_hash): Likewise.
(Fsecure_hash): List the SHA-3 algorithms in the docstring.
(syms_of_fns): Define Qsha3_224, Qsha3_256, Qsha3_384, and Qsha3_512.
* test/lisp/net/gnutls-tests.el (gnutls-tests-internal-macs-upcased):
Filter out the new SHA-3 algorithms since they are currently not
implemented in gnutls.
* test/src/fns-tests.el (test-secure-hash): Add test cases for the new
algorithms.
* doc/lispref/text.texi (Checksum/Hash): List the SHA-3 algorithms.
Mention that they are considered secure.
* etc/NEWS: Mention the new feature.
Users can revert to the old 'delete-frame' behavior of selecting
the oldest frame in 'frame-list' rather than the most-recently
used frame.
* src/frame.c (delete_frame): Consult
'after-delete-frame-select-mru-frame'.
(syms_of_frame): New defvar
'after-delete-frame-select-mru-frame'.
* etc/NEWS: Reflect the new user option.
* src/lisp.h (ENUM_BF): Modernize ENUM_BF doc for latest compilers
that still have a beef with enum bitfields. Simplify by dropping
support for ‘gcc -std=c89 -pedantic’, as it’s been many years
since Emacs worked with C89, for other reasons.
ISO C does not require C compilers to support enum bitfields.
Problem found with IBM XL C for AIX 16.1 (2018), which IBM
supports through 2029.
* src/lisp.h (struct Lisp_Fwd): Use ENUM_BF instead of declaring
enum bitfields directly.
In nsfns.m, `x-display-mm-height' and `x-display-mm-weight' computes the
display size by dividing its (logical) pixel dimensions by 92 dpi.
This would often give wrong results: by default, logical pixels on in
macOS are computed to 72 dpi, but that can be changed by the user.
As macOS multi-screen geometry is all computed based on the main
screen's cooridnate system, use its dpi to compute the physical size of
the bounding box containing all the screens.
* src/nsfns.m (Fx_display_mm_height):
(Fx_display_mm_width): Calculate the total physical display size
using the main-screen's geometry.
Copyright-paperwork-exempt: yes
This reverts commit 9f5d17cd72, but
includes the fix for adding user CSS after the generated CSS.
* src/image.c (svg_load_image): Set the CSS stylesheet using the librsvg
functions, instead of inserting it into the wrapper.
The keys of this plist are strings and should thus be compared
with `equal`. `eq` works surprisingly often (because we often
happen to use exactly the same string) but not always.
* lisp/loadhist.el (loadhist-unload-element):
* src/data.c (add_to_function_history): Pass `equal` to `plist-member`.
* lisp/frame.el (get-mru-frame): New defun.
* src/frame.c (delete_frame): Call 'get-mru-frame' (when force
is not Qnoelisp) to select the most recently used frame that is
not the deleted frame as the candidate to select.
(syms_of_frame): Qget_mru_frame new DEFSYM.
* doc/lispref/frames.texi: Document the new functions.
* etc/NEWS: Announce the new functions.
* src/treesit.c (ts_tree_cursor_copy) [WINDOWSNT]: Define and load
from the DLL if 'ts_tree_cursor_goto_previous_sibling' is not
available. This fixes the MS-Windows build broken by the last
change here. (Bug#80108)
Now we use treesit_traverse_sibling_helper when it exists, and
use the old code otherwise. So we still support older
tree-sitter versions.
* configure.ac (LIBSYSTEMD_CFLAGS): Add back old config checking
for the malloc function. Add another check for
ts_tree_cursor_goto_previous_sibling.
* src/treesit.c: (treesit_traverse_sibling_helper): Add back the
old code that doesn't require
ts_tree_cursor_goto_previous_sibling.
* src/xdisp.c (compute_display_string_pos): Pay attention to
'display' properties that come from overlays, in case the overlays
are window-specific. Call 'handle_display_prop' with the argument
OVERLAY non-nil, if the property is from an overlay.
(compute_display_string_end): Accept an additional argument, a
pointer to the window; all callers changed. Pay attention to
'display' properties that come from overlays, in case the overlays
are window-specific. In particular, if the property comes from an
overlay, use that overlay's end position as the next place where
the 'display' property changes, effectively disregarding any
properties on buffer text that is replaced by this overlay's
'display' property.
(handle_single_display_spec): Always use the end position of the
overlay which determined the 'display' property as the end of the
property, to make sure we obey windows-specific overlays.
* src/bidi.c (bidi_fetch_char): Adapt to the above change.
* src/dispextern.h: Adjust prototype of 'compute_display_string_end'.
(Bug#80255)
* src/buffer.c (overlays_in): Fix commentary to match the code.
(Foverlays_in, Foverlays_at): Doc fixes. (Bug#80242)
* test/src/buffer-tests.el (test-overlays-at-2): Add one test.
* doc/lispref/display.texi (Finding Overlays): Update and improve
the documentation of 'overlays-in' and 'overlays-at'.
* etc/NEWS: Announce the changes.
* src/pgtkfns.c (xg_set_icon_from_xpm_data):
* src/xfns.c (xg_set_icon_from_xpm_data):
gdk-pixbuf 2.44 (2025-09) deprecated xg_set_icon_from_xpm_data,
and this breaks --enable-gcc-warnings builds.
For now, pacify GCC by using a pragma.
Uses of <vla.h>’s macro VLA_ELEMS were removed some time ago.
* admin/merge-gnulib (GNULIB_MODULES): Remove vla.
* lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate.
* lib/vla.h: Remove.
* src/fns.c, src/lread.c: Do not include <vla.h>.
The greedy regexp matching, broken scoring and broken highlight were
sources of frequent complaints about the 'flex' matching style. This
commit fixes that.
Inspired by the 'hotfuzz' style (available at
https://github.com/axelf4/hotfuzz) it uses a modified version of Gotoh's
1982 dynamic programming algorithm. It is strictly more correct than the
"old" flex. For example, when matching the pattern 'goto' to no longer
will 'eglot-format' be sorted before some hypothetical much better
'goobarbaz-goto'. And of course the highlighting is also correctly
placed on the 'goto', not scattered across the candidate.
Regarding performance, it is faster than the naive 'flex', primarily
because of the Elisp rewrite in minibuffer.el. The matching and costing
algorithm matters but is not the bottleneck.
The Elisp parts of the style were almost completely decoupled from the
pcm/substring styles in lisp/minibuffer.el. Only
'completion-flex-try-completion' uses some of pcm's code for pattern
augmentation.
* src/minibuf.c (completion--flex-cost-gotoh): New function.
* lisp/minibuffer.el (completion-flex--pattern-str): New variable.
(flex-score-match-tightness): Make obsolete.
(completion--flex-all-completions-1): New helper function.
(completion-flex-try-completion, completion-flex-all-completions): Rewrite.
(completion-substring--all-completions): No longer take transform-pattern-fn.
(completion--flex-adjust-metadata): Tweak.
(completion--flex-score, completion--flex-score-1)
(completion--flex-score-last-md, completion-flex--make-flex-pattern): Delete.
* test/lisp/minibuffer-tests.el (completion--sorted-flex-completions):
New helper function.
(completion-flex-test-non-ascii): New test.
(completion--pcm-score): Delete.
(completion-pcm-test-3, completion-pcm-test-4)
(completion-substring-test-1, completion-substring-test-2)
(completion-flex-test-2, completion-flex-test-3): Tweak.
* etc/NEWS: Describe change.
