Rework the way we handle loops. This new code does not really work
better than the previous one, but it has the advantage of being "fail
safe" and also that we can dynamically check if our assumptions about
the shape of the bytecode are satisfied or not.
* src/regex-emacs.c (mutually_exclusive_aux): Replace `done_beg` and
`done_end` with `loop_beg` and `loop_end`.
(mutually_exclusive_p): Adjust accordingly.
(analyze_first): Fix incorrect assertion.
* src/xdisp.c (vadd_to_log): Fix off-by-1 typo that I introduced
in commit fbee6265a7
dated Thu Aug 13 13:48:28 2015 -0700.
Problem and fix reported by Robert Pluim (Bug#66098#11).
This makes for safer code when tagging null pointers in particular,
since pointer arithmetic on NULL is undefined and therefore can be
assumed, by the compiler, not to occur.
* src/lisp.h (untagged_ptr): Remove.
(TAG_PTR): Cast to uintptr_t instead of untagged_ptr.
* src/android.c (android_exception_check_1)
(android_exception_check_2, android_exception_check_3)
(android_exception_check_4): Permit each object to be NULL.
(android_browse_url): Release VALUE if an exception transpires.
* src/lisp.h (XUNTAG):
Instead of casting a Lisp value to char * and subtracting the tag,
cast it to a suitable integral type and work on that.
This should result in identical or at least equivalent code, except
that it avoids potential problems arising from the restrictions on
pointer arithmetic in C. In particular, a null pointer can be neither
an operand in nor the result of pointer arithmetic.
C compilers know this and would, prior to this change, optimise
XUNTAG(obj, Lisp_Int0, mytype) != NULL
to 1. This means, for example, that make_pointer_integer and
XFIXNUMPTR could not be entrusted with null pointers, and
next_vector in alloc.c was unsafe to use.
* java/org/gnu/emacs/EmacsDrawLine.java (perform):
* java/org/gnu/emacs/EmacsDrawRectangle.java (perform): Call
damageRect with integer coordinates in lieu of consing a new
Rect.
* java/org/gnu/emacs/EmacsDrawable.java (damageRect) <IIII>:
Declare a new variant of damageRect, accepting four integers
designating the extents of the damage rectangle.
* java/org/gnu/emacs/EmacsPixmap.java (damageRect) <IIII>: New
stub.
* java/org/gnu/emacs/EmacsView.java (damageRect) <IIII>:
Implement this overloaded variant of damageRect.
* java/org/gnu/emacs/EmacsWindow.java (damageRect): Ditto.
* src/android.c (android_init_emacs_drawable)
(android_init_emacs_window): Move search for `damageRect' to
android_init_emacs_window.
(android_damage_window): Call IIII variant of `damageRect' to
avoid consing a new rectangle. Ameliorate dynamic method
dispatch overhead.
This function should have been made internal in the first place.
* lisp/help.el (help--describe-map-tree): Rename from
'describe-map-tree'. Keep old name as an obsolete alias, and update
all callers.
* src/lisp.h (XUNTAG):
Instead of casting a Lisp value to char * and subtracting the tag,
cast it to a suitable integral type and work on that.
This should result in identical or at least equivalent code, except
that it avoids potential problems arising from the restrictions on
pointer arithmetic in C. In particular, a null pointer can be neither
an operand in nor the result of pointer arithmetic.
C compilers know this and would, prior to this change, optimise
XUNTAG(obj, Lisp_Int0, mytype) != NULL
to 1. This means, for example, that make_pointer_integer and
XFIXNUMPTR could not be entrusted with null pointers, and
next_vector in alloc.c was unsafe to use.
* src/alloc.c: The size of a vector block is bound by the number of
words, not bytes, represented by the pseudovector header RESTSIZE
field, because that limits how big a PVEC_FREE object can be.
755ae813a6 ; Declare some treesit.c functions in typescript-ts-mode.el.
89fa204b70 Fix loss of encrypted data in plstore.el
d9a1175a61 Close SQL database when corresponding 'sqlite-mode' buffe...
cbd8fac283 Fix Unicode normalization of characters
825be05b37 Support one-time passwords in Tramp
f880b94e64 Fix the 'C' and 'c' categories of characters
58fd212d8a Fix Emoji zooming commands
8970cdd009 ; Fix last change.
ba924be452 ; * etc/DEBUG: Improve the redisplay section.
e110312ad9 ; * doc/lispref/minibuf.texi (Text from Minibuffer): Ment...
65f4810003 tsx-ts-mode--font-lock-compatibility-bb1f97b: Improve
6fe11b88ed Avoid using --display in emacsclient to reuse frames on PGTK
2fc7463c0e ; * INSTALL: Don't advertise -O3. (Bug#65988)
29055412f2 ; Fix doc string of 'lsh'
738d854333 Support emacsclient on Windows with server on GNU or Unix...
f0a89fa1d0 ; * lisp/saveplace.el (save-place-ignore-files-regexp): F...
c9cb8ee0fc Fix defcustom in saveplace.el (Bug#65977)
5ec8be1d58 ; * lisp/subr.el (string-suffix-p, string-prefix-p): Doc ...
809305e6d8 Fix 'window-text-pixel-size' when there are several image...
ea14b0dcc2 : Doc fix.
01e8a0c6cb Doc fix for prettify-symbols-unprettify-at-point
0065621d0d (report_overlay_modification): Fix bug#65929
6cc6455e93 Fix SVG colors (bug#56182)
9396d73942 * doc/emacs/text.texi (Outline Minor Mode): Add a note ab...
a65d1a5a16 Improve documentation of 'list-abbrevs'
5dcc4b7eab Tweak s-p-f for js-ts-mode
1fb2fb501f typescript-ts-mode, tsx-ts-mode: Fix syntax properties fo...
946b395e7e * lisp/progmodes/c-ts-mode.el (c++-ts-mode): Provide (bug...
33ee3e588f Fix regression of treesit_cursor_helper_1
d11d81dfcc ; Fix doc typos (Bug#65868)
6554ec2246 Update docs for passing of Thien-Thi Nguyen
5ab2792d5c Update defvar usage tips example in manual
35d88c657e Document using Flymake together with Eglot
3f04efe9e7 ; * src/font.h (struct font): Comment about use of averag...
459b5f6b6d ; * admin/authors.el (authors-aliases): Update.
0c029ae8bc ; tweak etc/TODO item
# Conflicts:
# admin/authors.el
# lisp/subr.el
* java/org/gnu/emacs/EmacsContextMenu.java (display): Return
false if the list of menu buttons is empty, lest Android cease
displaying menus on the assumption that Emacs is defective.
* java/org/gnu/emacs/EmacsView.java (popupMenu): Likewise.
* src/fns.c (sort_list): Render sentence motion commands
functional within commentary
* src/sfntfont.c (sfntfont_list_family): Sort and deduplicate
the returned family list and make it a list of symbols.
(syms_of_sfntfont) <Qstring_lessp>: New defsym.
Commit 6fad73d7cc throws away some useful optimization because
it misfired in some cases (as seen in bug#657260). Here we try to
recover those useful optimizations with a slightly more careful
algorithm.
* src/regex-emacs.c (mutually_exclusive_aux): Rename from
`mutually_exclusive_p`. Add two new args. Improve the
case where we need to recurse.
(mutually_exclusive_p): New function defined on top of it.
* test/src/regex-emacs-tests.el (regexp-tests-backtrack-optimization):
Re-enable the test.
Minor refactoring to avoid swapping p1/p2.
* src/regex-emacs.c (mutually_exclusive_exactn)
(mutually_exclusive_charset): New functions, extracted from
`mutually_exclusive_p`.
(mutually_exclusive_p): Use them.
* src/process.c (child_signal_notify): Expunge call to
non-reentrant function from signal handler.
* src/xfns.c (setup_xi_event_mask): Cease selecting for
XI_FocusIn and XI_FocusOut under X toolkit builds.
* src/xterm.c (xi_handle_interaction, xi_handle_focus_change)
(xi_handle_focus_change): Make conditional on GTK 3 or no
toolkit builds.
(xi_handle_delete_frame, x_free_frame_resources): Adjust
correspondingly.
(handle_one_xevent) <EnterNotify, LeaveNotify>: Examine
EnterNotify and LeaveNotify events for focus changes
irrespective of whether XI2 is enabled under the X toolkit and
GTK 2.x.
* java/org/gnu/emacs/EmacsFontDriver.java (FONT_INVALID_CODE):
New constant.
(hasChar, encodeChar): Accept a jint character code rather than
a jchar.
* java/org/gnu/emacs/EmacsSdk23FontDriver.java (hasChar)
(encodeChar): Return 0 or FONT_INVALID_CODE if the character is
not representable in a Java char.
* java/org/gnu/emacs/EmacsSdk7FontDriver.java (hasChar): Ditto.
* src/androidfont.c (android_init_font_driver): Adjust signature
to match.
Somehow the new overlay code in Emacs-29 changed slightly
the test to decide when to run the `modification-hooks` of overlays,
with the consequence that these hook functions end up being executed
also when text is deleted right after an empty overlay, which is
contrary to Emacs-28 behavior as well as contrary to the Texinfo doc.
* src/buffer.c (report_overlay_modification): Better reproduce the
Emacs-28 code.
* test/src/buffer-tests.el (overlay-modification-hooks):
Add corresponding test.
* src/treesit.h (Lisp_TS_Parser)
* src/treesit.c (make_treesit_parser): Remove has_range field, add
last_set_ranges field.
(Ftreesit_parser_set_included_ranges): Return early if the new range
is the same as the old one.
(Ftreesit_parser_included_ranges): Change has_range to
last_set_ranges.
A previous change should've achived this, but that wasn't enough.
This one actually achieves the intended effect, namely,
treesit-node-match-p doesn't raise an error if IGNORE-MISSING is
non-nil, and traverse functions don't raise error by default.
* src/treesit.c (treesit_traverse_validate_predicate): Remove
ignore_missing parameter, return error symbol.
(Ftreesit_node_match_p)
(Ftreesit_search_subtree)
(Ftreesit_search_forward)
(Ftreesit_induce_sparse_tree): Check returned error symbol.
(syms_of_treesit): New symbol.
* src/sfnt.c (sfnt_table_names): Append an entry for OS/2
tables.
(sfnt_read_OS_2_table): New function.
(main): Introduce a new test for the OS/2 table reader.
* src/sfnt.h (sfnt_read_OS_2_table): New declaration.
* src/sfntfont.c (sfnt_decode_foundry_name): Delete function.
(sfnt_enum_font_1): Read the font's OS/2 table and extract the
foundry name from there. Use `misc' if absent.
(sfntfont_desc_to_entity, sfntfont_open): Cease interning
desc->designer, as that is now a symbol.
(syms_of_sfntfont) <Qmisc>: New defsym Qmisc. (bug#65865)
* src/treesit.c (treesit_cursor_helper_1)
(treesit_cursor_first_child_for_byte): Use
ts_tree_cursor_goto_first_child_for_byte first, and
ts_tree_cursor_goto_first_child second.
It seems this was likely a bug in macOS 10.10.1. I have no way to find
out when it was fixed, but hopefully before 10.10.5, the most recent
release of that version.
* src/nsterm.m (ns_send_appdefined): Remove check for whether previous
appdefined events have been lost.
* src/nsterm.m ([EmacsView initFrameFromEmacs:]): Reorder the way the
frame and layers are created.
([EmacsView makeBackingLayer]): Change to the newly renamed method
below.
([EmacsLayer initWithColorSpace:doubleBuffered:]):
([EmacsLayer initWithDoubleBuffered:]): Rename the method and remove
the colorspace argument as it's no longer able to be set on initial
creation.
* src/nsterm.h: Use new method prototype.
