* src/lisp.h (TAG_PTR_INITIALLY): Rename from TAG_PTR, since calls
can be used only as initializers, and the convention elsewhere in
lisp.c is to give these macros names ending in ‘_INITIALLY’.
This should help avoid confusion such as we recently experienced
in make_lisp_symbol_internal. All uses changed.
* src/lisp.h (LISPSYM_INITIALLY): Remove unnecessary cast to char *.
The C standard allows but does not require support for casts to
pointers in constant expressions in static initializers.
* src/w32term.c (w32_flip_buffers_if_dirty): Do nothing if F is
not a GUI frame. This avoids rare crashes in "emacs -nw".
* src/w32console.c (initialize_w32_display): Set the
ENABLE_EXTENDED_FLAGS bit in 'prev_console_mode'.
* src/w32fns.c (globals_of_w32fns) <w32-follow-system-dark-mode>:
New variable.
(w32_applytheme): Disable application of Dark mode if
'w32-follow-system-dark-mode' is nil.
* etc/NEWS:
* doc/emacs/msdos.texi (Windows Misc): Document
'w32-follow-system-dark-mode'.
Revert DOHASH to the fast (field-caching) implementation but with
an assertion to detect misuses. Add DOHASH_SAFE for use in
code that must tolerate arbitrary mutation of the table being
iterated through.
* src/lisp.h (DOHASH): Go back to fast design that only allows
restricted mutation, but with a checking assertion.
(DOHASH_SAFE): New macro that tolerates arbitrary mutation while being
much simpler (and acceptably fast).
* src/fns.c (Fmaphash):
* src/comp.c (compile_function, Fcomp__compile_ctxt_to_file):
Use DOHASH_SAFE.
The problem was that the offset computed for the `charset_table` array
was a multiple of 4 but the `struct charset` needed an alignment on
a multiple of 8, so `dump_charset` inserted 4 bytes of padding,
whereas you can't have padding at the beginning of an array.
* src/pdumper.c (dump_charset): Don't set alignment here.
(dump_charset_table): Set it here instead.
* src/w32fns.c (DEFAULT_IMAGE_BASE): Define for 64-bit and 32-bit
MinGW builds.
(emacs_abort): Correct the callstack addresses for potential
relocation of the image base due to ASLR. This makes 'addr2line'
be able to interpret emacs_backtrace.txt when ASLR is in effect,
which it is on every modern version of MS-Windows. (Bug#63365)
* configure.ac (LD_SWITCH_SYSTEM_TEMACS) [mingw32]: Add comment
about keeping the image-base values in sync with w32fns.c.
* etc/DEBUG (How to disable ASLR): New section.
* java/org/gnu/emacs/EmacsView.java (swapBuffers): Synchronize
such that code cannot execute between the bitmap's being loaded
and being transferred to surfaceView.
(onDetachedFromWindow): Recycle bitmap after the surface view is
reset.
* java/org/gnu/emacs/EmacsWindow.java (recreateActivity):
* src/android.c (android_init_emacs_window)
(android_recreate_activity):
* src/androidfns.c (Fandroid_recreate_activity)
(syms_of_androidfns): New functions for debugging window
attachment.
* src/androidgui.h: Update prototypes.
* src/androidfns.c (android_create_tip_frame): Enable building
with GLYPH_DEBUG.
* src/window.c (resize_mini_window_apply): Garbage the frame if
F->redisplay is already set to indicate that redisplay_internal
should nevertheless return to it.
I gave too much credit to the comment, and didn't realize that macro
was used in places that didn't obey the comment.
This macro is getting pretty hideous!
`struct composition` kept an index into the internal `key_and_value` array
of hash tables, which only worked because of details of how
hash-tables are handled. Replace it with a reference to the
key stored at that location in the hash-table, which saves us an
indirection while at it.
* src/composite.h (struct composition): Replace `hash_index` with
the actual `key`.
(COMPOSITION_KEY): Simplify accordingly.
(mark_composite): Declare.
* src/composite.c (get_composition_id): Adjust accordingly.
(mark_composite): New function.
* src/charset.c (mark_charset): Uncomment.
* src/lisp.h (mark_charset): Declare.
* src/alloc.c (garbage_collect): Call `mark_charset` and `mark_composite`.
* src/pdumper.c (hash_table_contents): Remove invalid comment, since
compositions aren't dumped.
This leads to simpler code in the users, and more efficient machine
code because we don't repeatedly need to fetch the `table_size`
and `key_and_value` fields of the hash table object.
* src/lisp.h (DOHASH): Rewrite.
* src/composite.c (composition_gstring_lookup_cache): Simplify.
(composition_gstring_cache_clear_font):
* src/print.c (print):
* src/pdumper.c (hash_table_contents):
* src/minibuf.c (Ftest_completion):
* src/json.c (lisp_to_json_nonscalar_1):
* src/emacs-module.c (module_global_reference_p):
* src/comp.c (compile_function, Fcomp__compile_ctxt_to_file):
* src/fns.c (Fmaphash): Adjust to new calling convention.
Simplify optimization for make_lisp_symbol, so that
it’s less tricky and works well enough for gcc -Og or -O2.
* src/lisp.h (lisp_h_builtin_lisp_symbol): Remove.
(builtin_lisp_symbol) [DEFINE_KEY_OPS_AS_MACROS]: Remove.
(make_lisp_symbol_nodebug): New internal static function,
which is like the old make_lisp_symbol but without the eassert.
(make_lisp_symbol, builtin_lisp_symbol): Use it, so that
make_lisp_symbol has the eassert but builtin_lisp_symbol doesn’t.
Co-authored-by: Paul Eggert <eggert@cs.ucla.edu>
`struct charset` kept an index into the internal `key_and_value` array
of hash tables, which only worked because of details of how
hash-tables are handled. Replace it with a reference to the
value stored at that location in the hash-table, which saves us an
indirection while at it.
* src/charset.h (struct charset): Replace `hash_index` field with
`attributes` field.
(CHARSET_ATTRIBUTES): Simplify accordingly.
(CHARSET_HASH_INDEX): Delete unused macro.
* src/charset.c (Fdefine_charset_internal):
* src/pdumper.c (dump_charset): Adjust accordingly.
(dump_charset_table): Set the referrer since that's needed while
dumping Lisp_Object fields.
* src/sfnt.c (sfnt_address_zp2, sfnt_address_zp1)
(sfnt_address_zp0): Don't save into X or Y if the zone is set to
the twilight zone and they are NULL.
* src/sfnt.c (sfnt_mul_f26dot6_round): New function.
(sfnt_mul_f26dot6_fixed): Replace by call to
sfnt_mul_fixed_round.
(MUL): Round result, as the Apple and MS scalers do.
(sfnt_interpret_control_value_program): The instruction control
flag which reverts CVT modifications is 2, not 4.
Problem reported by Alan Mackenzie in:
https://lists.gnu.org/r/emacs-devel/2024-01/msg00755.html
* src/lisp.h (XSYMBOL): If the arg is not a bare symbol, then
eassert (symbols_with_pos_enabled). This shouldn’t affect code
generated for regular builds, and could catch caller errors in
debug builds. For debug builds although this slows things down
XSYMBOL should still be faster than it was the day before
yesterday, as there’s still no need to eassert (SYMBOLP (a)).
This should help when building with --enable-checking and
compiling with gcc -O0. Problem reorted by Stefan Monnier in:
https://lists.gnu.org/r/emacs-devel/2024-01/msg00770.html
* src/lisp.h (lisp_h_builtin_lisp_symbol): New macro,
with a body equivalent in effect to the old ‘builtin_lisp_symbol’
but faster when not optimizing.
(builtin_lisp_symbol): Use it.
If DEFINE_KEY_OPS_AS_MACROS, also define as macro.
This removes hacks from code that had to be careful not to use
Qunbound as a hash table key, at the cost of a minor hack in
the GC marker.
* src/lisp.h (INVALID_LISP_VALUE, HASH_UNUSED_ENTRY_KEY):
Define as a null-pointer float.
* src/alloc.c (process_mark_stack): Add hack to ignore that value.
* src/pdumper.c (dump_object_needs_dumping_p)
(pdumper_init_symbol_unbound, pdumper_load):
* src/print.c (PRINT_CIRCLE_CANDIDATE_P): Remove hacks for Qunbound.
Recent hash table changes reduced the range of sxhash, sxhash-eq etc
to [0,2**32) on platforms with 62-bit fixnums. This change makes them
use the full fixnum range again. Hash table hashing is unaffected.
* src/fns.c (sxhash_eq, sxhash_eql): New.
(hash_hash_to_fixnum): Replace with...
(reduce_emacs_uint_to_fixnum): ...this.
(hashfn_eq, hashfn_eql, Fsxhash_eq, Fsxhash_eql, Fsxhash_equal)
(Fsxhash_equal_including_properties): Use the new functions.
* doc/lispref/hash.texi (Hash Access):
* src/fns.c (Fmaphash):
Make it clear what the function passed as argument can do. Until now
these rules were unwritten, and are still unenforced.
* src/lisp.h (make_lisp_symbol): In eassert use XBARE_SYMBOL
rather than XSYMBOL. This is safe because the symbol must be
bare. The change speeds up make_lisp_symbol when debugging.
* src/lisp.h (XSYMBOL): Simplify and tune. There is no need to
examine symbols_with_pos_enabled here, since the arg must be a symbol
so if it's not a bare symbol then it must be a symbol_with_pos;
and checking whether a symbol is bare is cheap.
With Ubuntu 23.10 on a Xeon W-1350, this shrank Emacs’s executable
text size by 0.1% and sped up a default build of all *.elc files by
0.4%.
Remove unnecessary eassert, since XBARE_SYMBOL and XSYMBOL_WITH_POS
have easserts that suffice.
Be more systematic about putting space before paren in calls,
and in avoiding unnecessary parentheses in macros.
This was partly inspired by my wading through gcc -E output
while debugging something else, and seeing too many parens.
This patch does not change the generated .o files on my platform.
* src/lisp.h (XBARE_SYMBOL, XSYMBOL): Omit parentheses that are no
longer needed now that we have symbols with positions and these
symbols are never macros.
5bb5590dec Fix blunder in labeled_narrow_to_region
78ddb32fad Fix documentation of icon-elements
725a3f32f8 ; Fix typos in symbol names
6653ee66ca Improve two docstrings in ox-latex
7d869a0402 Doc fix in auth-source-read-char-choice
f149de223b Merge branch 'emacs-29' of git.savannah.gnu.org:/srv/git/...
1f97a87879 Fix info-xref-tests
51f391998b Add @kindex in manuals for existing keybindings on 'C-x x...
* src/sfnt.c (sfnt_multiply_divide_rounded): New function.
(sfnt_multiply_divide_signed): Always round values, as fonts
which rely on IP to move points in concert with prior motion and
subsequently round such points with MDAP are sensitive to minor
deviations in the behavior of the former instruction.
(load_unscaled): New macro.
(IUP_SINGLE_PAIR, sfnt_interpret_iup_1): Compute ratio w/
unscaled points if possible.
This flaw could cause an assertion failure.
* src/print.c (PRINT_CIRCLE_CANDIDATE_P): Don't consider Qunbound a
print-circle candidate; it should never be seen by Lisp anyway.
This invariant was intended but insufficiently enforced which could
lead to an assertion failure.
* src/fns.c (hash_index_size): Assume size>0, and return a value >1.
(make_hash_table): Only use hash_index_size for size>0.
* src/fns.c (hash_table_thaw): Don't allocate anything for empty
tables. Don't initialise the next vector twice.
(maybe_resize_hash_table): Factor out min_size constant.
* src/editfns.c (labeled_narrow_to_region): Record point before,
instead of after, calling narrow-to-region; otherwise point may
already have been changed. Fixes bug#66764.
