* src/alloc.c (mark_maybe_object, mark_maybe_objects): Remove.
(mark_objects): New function.
* src/eval.c (mark_specpdl): Use mark_objects instead of
mark_maybe_objects, since the array now has only valid Lisp objects.
* src/lisp.h (SAFE_ALLOCA_LISP_EXTRA): When allocating a large
array, clear it so that it contains only valid Lisp objects. This
is simpler and safer, and does not hurt performance significantly
on my usual benchmark as the code is executed so rarely.
This lets Emacs avoid marking some garbage as if it were in use.
On one test platform (RHEL 7.8, Intel Xeon Silver 4116) it
sped up ‘cd lisp; make compile-always’ by a bit over 1%.
* src/alloc.c (live_string_holding, live_cons_holding)
(live_symbol_holding, live_large_vector_holding)
(live_small_vector_holding):
Count only pointers that point to a struct component,
or are a tagged pointer to the start of the struct.
Exception: for non-bool-vector pseudovectors,
count any pointer past the header, since it’s too much
of a pain to write code for every pseudovector.
(live_vector_pointer): New function.
* src/alloc.c (mark_memory): Do not bother using mark_maybe_object
on the stack, since mark_maybe_pointer now marks everything that
mark_maybe_object would.
On --with-wide-int platforms where Lisp_Object can be
put into non-adjacent registers, mark_maybe_pointer failed
to mark a float whose only reference was as a tagged pointer.
* src/alloc.c (live_float_holding): New function,
a generalization of the old live_float_p.
(live_float_p): Use it.
(mark_maybe_pointer): Use live_float_holding, not live_float_p.
* src/fileio.c (Fexpand_file_name): Restore pre-August-26
behavior, if DOS_NT. This should fix the recently-introduced
expand-file-name bugs on DOS_NT (Bug#26911).
The bug was that (expand-file-name "~") returned something
like "/home/eggert/" instead of "/home/eggert".
Problem reported by Mattias Engdegård (Bug#26911#27).
* src/fileio.c (Fexpand_file_name): When concatenating NEWDIR to
NM, instead of stripping trailing slashes from NEWDIR (which can
turn non-symlinks into symlinks), strip leading slashes from NM.
This also simplifies the code by removing no-longer-needed DOS_NT
special-casing. Also, remove an unnecessary ‘target[length] = 0;’
as that byte will be overwritten by the next memcpy anyway.
* test/src/fileio-tests.el (fileio-tests--HOME-trailing-slash):
New test.
Problem reported by Yegor Timoshenko (Bug#26911),
and I ran into it myself recently in normal-top-level.
* doc/lispref/files.texi (File Name Expansion), etc/NEWS: Mention this.
* src/fileio.c (Fexpand_file_name): Expand "/a/b/." to "/a/b/" not
"/a/b", to avoid misinterpreting a symlink "/a/b". Similarly,
expand "/a/b/c/.." to "/a/b/" not "/a/b".
* test/lisp/net/tramp-tests.el (tramp-test05-expand-file-name):
Adjust to match new behavior.
(tramp-test05-expand-file-name-relative): This test now succeeds,
at least on Fedora 31.
* test/src/fileio-tests.el:
(fileio-tests--expand-file-name-trailing-slash) New test.
* src/regex-emacs.c (PUSH_FAILURE_REG, POP_FAILURE_REG_OR_COUNT)
(re_match_2_internal): Add some assertions that regstart
is set whenever regend is.
(re_match_2_internal): Omit two unnecessary REG_UNSET (regstart ...)s
and one unnecessary assignment to regend.
Problem noted by Stefan Monnier in:
https://lists.gnu.org/r/emacs-devel/2020-08/msg00846.html
* src/font.c (xft_ignore_color_fonts):
* src/syntax.c (comment_end_can_be_escaped):
* src/xdisp.c (word_wrap_by_category, display_fill_column_indicator):
Rename these DEFVAR_BOOL variables to avoid the initial "V"
that wrongly suggests that they are Lisp_Object variables.
All uses changed.
* src/keymap.c (describe_map): A binding may be shadowed by
something else than a mode (bug#14086) (just a `define-key'
works), so don't say that it's a mode that shadows it.
* src/editfns.c (NOTE_DELETE, NOTE_INSERT): Avoid unnecessary parens.
(Freplace_buffer_contents): Check args before returning results.
Avoid integer overflow when computing too_expensive, and work even
if MAX-COSTS is bignum. Call alloca and/or malloc just once, not
three times.
(set_bit, bit_is_set): Simplify micro-optimization by using eassume.
* src/editfns.c (rbc_quitcounter): Remove; the quitcounter
is now part of the context.
(EXTRA_CONTEXT_FIELDS): Remove unused member early_abort_tests.
Add jmp, quitcounter.
(Freplace_buffer_contents): Use setjmp/longjmp to recover from
a compareseq that runs too long. Omit unnecessary rarely_quit
call.
(buffer_chars_equal): Occasionally check for early abort and
longjmp out if so (Bug#43016).
* src/gnutls.c (gnutls_symmetric): Cache the results from
Fgnutls_ciphers, since that function isn't very fast (bug#42998).
(syms_of_gnutls): Initialize cache variable.
Inspired by a straightforward patch by Bruno Haible.
* admin/merge-gnulib (GNULIB_MODULES): Add sigdescr_np.
* configure.ac: Do not check for sys_siglist or __sys_siglist.
* lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate.
* lib/sigdescr_np.c, m4/sigdescr_np.m4: New files, copied from Gnulib.
* src/sysdep.c (sys_siglist, sys_siglist_entries): Remove.
(init_signals): Do not initialize sys_siglist.
(safe_strsignal): Use sigdescr_np instead of sys_siglist.
This fixes assertion violations when the mnemonic is
given as a string, and allows non-ASCII characters be
used as mode-line mnemonic of a coding-system.
* src/xdisp.c (decode_mode_spec_coding): Handle multibyte
characters as coding-system's mnemonic.
(display_mode_element): If decode_mode_spec returns a multibyte
string, display it as multibyte.
* src/coding.c (Fdefine_coding_system_internal)
(Fcoding_system_put): If :mnemonic is a string, use its first
character. This avoids assertion violations if someone uses a
string as the mnemonic of a coding-system.
* src/xdisp.c (gui_consider_frame_title): Multibyte-encode any raw
bytes in the title, and then pass a multibyte string to the back-end
for use as a frame title. This cuts down a little on the rubbish
shown when raw bytes sneak in by mistake (as part of the buffer name,
for instance).
The new code calls 'malloc' and 'free', so we can no longer
* src/image.c (struct image_type): Rename 'load' to 'load_img' and
'free' to 'free_img'. All callers changed.
(free_image) [WINDOWSNT]: Don't #undef 'free'.
* test/manual/image-transforms-tests.el: Replace hard-coded colors
with defaults.
* src/dispextern.h (struct image):
* src/image.c (search_image_cache):
(xbm_load_image):
(xbm_load):
(pbm_load): Rename from frame to face where relevant.
(svg_load_image): Parse the image to find out the size, then wrap it
in another SVG to set a new size and colors, etc.
(lookup_image): Use the face colors instead of the frame colors.
(search_image_cache): Add ability to ignore the face colors.
(uncache_image): Uncache all copies of the image that share the spec,
even if the face colors don't match.
* etc/NEWS: Describe the changes.
Note about the change around line 9257 and 23372:
Before, the test for whitespace checks for can_wrap_before and
can_wrap_after simutaniously. Now we separate these two checks, and
the logic needs to change a little bit. However, when we don't enable
the new wrapping feature, 'can_wrap_after' is equivalent to
'IT_DISPLAYING_WHITESPACE' and 'can_wrap_before' is equivalent to
'!IT_DISPLAYING_WHITESPACE'. And the new logic is equivalent with the
old one in that case.
Old logic:
if (whitespace) /* Which means can wrap after && can't wrap
before. */
may_wrap = true;
else if (may_wrap) /* aka (!whitespace && may_wrap)
(set wrap point) * aka (can't wrap after && can wrap before
may_wrap = false * && may_wrap)
*/
New logic:
if (can_wrap_after)
next_may_wrap = true
else
next_may_wrap = false;
if (may_wrap && can_wrap_before)
(set wrap point)
/* Update may_wrap. */
may_wrap = next_may_wrap;
* src/xdisp.c (it_char_has_category, char_can_wrap_before)
(char_can_wrap_after): New functions.
(move_it_in_display_line_to, display_line): Replace calls to
'IT_DISPLAYING_WHITESPACE' with either 'char_can_wrap_before' or
'char_can_wrap_after'.
(word-wrap-by-category): New variable.
* lisp/cus-start.el (minibuffer-prompt-properties--setter): Add
'word-wrap-by-category' as a customizable variable.
* doc/emacs/display.texi (Visual Line Mode): Add a paragraph about the
new 'word-wrap-by-category' feature.
* etc/NEWS: Announce the change.
Instead of blindly assuming that all Emacs strings are valid UTF-8,
which they are not, use a more careful conversion going via UTF-16
which is what NSString uses internally. Unpaired surrogates will
still go through to the NSString objects, but the NS libs handle them
gracefully.
* src/nsterm.h (EmacsString): New category.
* src/nsfns.m (all_nonzero_ascii): New helper function.
([NSString stringWithLispString:]): New method.
(ns_set_name_internal): Use new conversion method.
* src/minibuf.c (read_minibuf_noninteractive): Signal an
`end-of-file' error when reading from stdin instead of a general
error (bug#34123). This makes it easier to write code that
recovers from this situation.
Suggested by Noam Postavsky <npostavs@gmail.com>.
* test/src/image-tests.el (image-test-circular-specs): New file.
* src/image.c (parse_image_spec): Return failure for circular lists.
(valid_image_p): Don't look at odd-numbered list elements expecting to
find a property name.
(image_spec_value): Handle circular lists.
(equal_lists): Introduce.
(search_image_cache): Use `equal_lists' (bug#36403).
362ca83a3b (origin/emacs-27) Let Emacs start even if curdir is inacce...
dd989c0ea0 * etc/NEWS: Mention GnuPG 2.0 through 2.1.5 issue (Bug#428...
4542b750cc Fix bug with ~/Emacs file not being read at init
9b403d624e ; Fix last change
6bff65a626 ; * doc/lispref/sequences.texi (Sequence Functions): Typo ...
3c4edfd85e Prevent from frozen frame after `C-z' in Lucid builds
98e8241992 Document the 'flex' completion style
19fa8b7ca3 Note that Emacs needs systemd support if systemd is used t...
# Conflicts:
# etc/NEWS
