* src/lisp.h: Remove include of emacs-module.h. Remove definition
of Lisp_Module_Function structure.
* src/emacs-module.c (module_function_documentation)
(module_function_address): New accessor functions for module function
fields.
(emacs_subr, struct Lisp_Module_Function): Move from lisp.h.
* src/print.c (print_vectorlike):
* src/doc.c (Fdocumentation): Use the new accessor functions.
* src/lisp.h (PVECHEADERSIZE): New macro.
(XSETPVECTYPESIZE): Use it.
* src/search.c (syms_of_search): No need to initialize or
staticpro last_thing_searched or saved_last_thing_searched, as
the thread code arranges for initialization and GC.
* src/thread.c (main_thread): Initialize statically.
(Fmake_mutex, Fmake_condition_variable, Fmake_thread):
Use ALLOCATE_ZEROED_PSEUDOVECTOR rather than zeroing by hand.
(mark_one_thread): No need to mark Lisp_Object members.
(init_main_thread, init_threads_once): Remove. All uses removed.
* src/xdisp.c (pos_visible_p): Account for line-number display
width when CHARPOS is covered by display property that begins
at the 2nd display element of a screen line. (Bug#35404)
make_time is documented to return a (TICKS . HZ) pair, so we can’t use
make_lisp_time. Introduce a new conversion function instead.
* src/emacs-module.c (module_make_time): Use timespec_to_lisp to
correct return type.
* src/timefns.c (timespec_to_lisp): New function.
(make_lisp_time): Use it.
* test/src/emacs-module-tests.el (mod-test-add-nanosecond/valid):
Check return type.
* src/module-env-27.h: Add new module functions to convert big
integers.
* src/emacs-module.h.in (emacs_mpz): Define if GMP is available.
* src/emacs-module.c (module_extract_big_integer)
(module_make_big_integer): New functions.
(initialize_environment): Use them.
* test/data/emacs-module/mod-test.c (Fmod_test_double): New test
function.
(emacs_module_init): Define it.
* test/src/emacs-module-tests.el (mod-test-double): New unit test.
* doc/lispref/internals.texi (Module Values): Document new functions.
Time values are a fundamental data type, and such conversions are hard
to implement within modules because of the various forms of time
values in Emacs Lisp. Adding dedicated conversion functions can
significantly simplify module code dealing with times.
This approach uses nanosecond precision. While Emacs in theory has
support for higher-precision time values, in practice most languages
and standards, such as POSIX, C, Java, and Go, have settled on
nanosecond-precision integers to represent time.
* src/emacs-module.h.in: Add header for struct timespec.
* src/module-env-27.h: Add module functions for time conversion.
* src/emacs-module.c (module_extract_time, module_make_time): New
functions.
(initialize_environment): Use them.
* test/data/emacs-module/mod-test.c (Fmod_test_add_nanosecond): New
test function.
(emacs_module_init): Define it.
* test/src/emacs-module-tests.el (mod-test-add-nanosecond/valid)
(mod-test-add-nanosecond/nil, mod-test-add-nanosecond/invalid): New
unit tests.
* doc/lispref/internals.texi (Module Values): Document time
conversion functions.
This reverts commit 64d0cd9810.
Rationale: any font-related code and comments, even if unused
for decades, serves as important source of useful information
in an area of Emacs code that is notoriously under-documented.
Please do NOT remove this stuff until we have an active
expert in this are on board, who will then decide whether
this can be retired.
These directives are in files that are compiled only if the
symbols are defined.
* src/gfilenotify.c: Remove unnecessary ‘#ifdef HAVE_GFILENOTIFY’.
* src/inotify.c: Remove unnecessary ‘#ifdef HAVE_INOTIFY’.
* src/kqueue.c: Remove unnecessary ‘#ifdef HAVE_KQUEUE’.
* src/font.c (LSTRING_HEADER_SIZE, LSTRING_GLYPH_SIZE, check_gstring)
(check_otf_features, otf_list, otf_tag_symbol, otf_open)
(font_otf_capability, generate_otf_features)
(font_otf_DeviceTable, font_otf_ValueRecord)
(font_otf_Anchor, Ffont_drive_otf, Ffont_otf_alternates)
(Fdraw_string, syms_of_font): Remove "experimental and not
tested much" code that has been "#if 0"-ed out for more than a
decade and which was getting in the way of maintenance.
INT_TO_INTEGER is more obviously correct and means we don’t have to
worry about data type sizes and signedness.
* src/json.c (json_parse_error): Use INT_TO_INTEGER. The tiny
performance gain of make_fixed_natnum isn’t worth the trouble then
signaling an error.
Thanks to Dmitry Gutov <dgutov@yandex.ru> for running many
benchmarks and for useful discussions.
* src/json.c (json_make_string): Speed up parsing of JSON
strings by optimizing the normal case of a valid UTF-8 string
being returned from libjansson. (Bug#31138)
Stefan Monnier pointed out examples like (funcall `(closure
,(let ((cycle (list nil))) (setcdr cycle cycle)) () a)),
where the user can set Vinternal_interpreter_environment
indirectly.
* src/eval.c (Fsetq): Revert recent change, going back to Fassq.
Although the lisp.h macros really need improvement,
INTEGER_TO_INT is not the right way to go about it, as it
causes conversion from intmax_t to uintmax_t and back again,
which can cause a signal if the value is negative.
* src/lisp.h (INTEGER_TO_INT, ranged_integer_to_int)
(ranged_integer_to_uint): Remove, reverting recent changes to
this file.
* src/json.c (lisp_to_json): Revert to previous code,
as the change messes up with uintmax_t<->intmax_t conversion.
This is similar to CONS_TO_INTEGER. The inverse (INT_TO_INTEGER)
already exists.
* src/lisp.h (INTEGER_TO_INT): New macro.
(ranged_integer_to_int, ranged_integer_to_uint): New
functions.
* src/json.c (lisp_to_json): Use helper macro.
* emacs-module.c (module_make_string): Use make_unibyte_string, which
doesn’t require its argument to be null-terminated. Since it always
returns a heap-allocated string, we don’t have to copy it any more
while decoding.
(module_decode): New helper function.
Factor out conversions into helper functions to provide a simpler
interface.
* src/emacs-module.c (module_encode, module_decode_copy): New helper
functions.
(module_make_function, module_copy_string_contents)
(module_make_string): Use them.
Now that CATCHER_ALL catches signals as well, we can simplify
MODULE_HANDLE_NONLOCAL_EXIT a bit.
* src/emacs-module.c (MODULE_SETJMP, MODULE_SETJMP_1): Remove.
(MODULE_HANDLE_NONLOCAL_EXIT): Inline MODULE_SETJMP and
MODULE_SETJMP_1.
It doesn’t work anymore, and these days ‘gcc -fsanitize=address’
does a better job anyway.
* configure.ac: Remove the option.
* configure.ac (ac_xmalloc_overrun, XMALLOC_OVERRUN_CHECK):
* src/alloc.c (XMALLOC_OVERRUN_CHECK_OVERHEAD)
(XMALLOC_OVERRUN_CHECK_SIZE, XMALLOC_OVERRUN_SIZE_SIZE)
(xmalloc_overrun_check_header, xmalloc_overrun_check_trailer)
(xmalloc_put_size, xmalloc_get_size, overrun_check_malloc)
(overrun_check_realloc, overrun_check_free):
Remove. All uses removed.
* etc/NEWS: Mention this.
Because XPNTR now uses ATTRIBUTE_NO_SANITIZE_UNDEFINED,
it can be simplified.
* src/alloc.c (macro_PNTR_ADD, PNTR_ADD, macro_XPNTR): Remove.
(XPNTR): Open-code rather than using the removed macros and
functions. Also, simplify by using LISP_WORD_TAG.
* src/pdumper.c (dump_charset): Use alignof (struct charset),
not alignof (int), since struct charset might be more strictly
aligned than int. I think this is just a minor performance
issue, but we might as well use the correct alignment.
(drain_reloc_list): Use an alignment instead of a size for the
output alignment. This prevents undefined behavior when
alignof (struct emacs_reloc) == 8 and sizeof (dump_off) == 4
when building on x86-64 with gcc -fsanitize=undefined.
* src/pdumper.c (dump_bitset_clear):
Pacify -fsanitize=undefined by avoiding memset (NULL, x, 0),
which strictly speaking has undefined behavior although it
works on all production platforms I know.
* src/alloc.c (XPNTR): Add ATTRIBUTE_NO_SANITIZE_UNDEFINED and
remove ATTRIBUTE_UNUSED. Do not define as a macro, so that
ATTRIBUTE_NO_SANITIZE_UNDEFINED works.
* src/lisp.h (lisp_h_XSYMBOL): Remove. All uses removed.
With recent GCC the macro does not work with -fsanitize=undefined,
and the macro can be omitted as its only function is to optimize -O0.
Revert the double-free bug that I introduced in
2019-03-11T15:20:54Z!eggert@cs.ucla.edu.
* src/pdumper.c (dump_mmap_reset): Do not free the private member;
that’s the release function’s job.
(dump_mm_heap_cb_release): Free cb if its refcount goes to zero.
(dump_mmap_contiguous_heap): Mention memory leak in comment.
Since DEFSYM doesn't by itself do anything and make-docfile ignores
preprocessor statements, conditional compilation of DEFSYMs is
ineffective.
* src/data.c (syms_of_data): Remove ineffective #ifdefs.
Even if Emacs is compiled without module support, we don't have to
comment out every bit of user pointer support. Defining the basic
structures and functions and detecting user pointers in switch
statements is harmless, and we're already doing the same for module
functions. Removing these #ifdefs makes the code a bit easier to
read.
* src/lisp.h (PVEC_USER_PTR, struct Lisp_User_Ptr, USER_PTRP)
(XUSER_PTR): Define unconditionally.
* src/data.c (Ftype_of):
* src/alloc.c (cleanup_vector):
* src/print.c (print_vectorlike):
* src/pdumper.c (dump_vectorlike): Remove #ifdef for user pointers.
This partially reverts commit
09b2b8a5ce.
* src/alloc.c (mark_maybe_pointer): Remove special-casing of tagged
pointers. After commit 09d746dad3,
modules no longer rely on tagged pointers.
* src/eval.c (eval_sub): Check whether Fassq returns Qnil,
not whether it returns a cons, as NILP is faster than CONSP
nowadays. Remove incorrect comment “only original_fun and
original_args have values that will be used below”; instead,
move declarations around so that the set of variables with
useful values is obvious.
See Bug#31676.
* lisp/emacs-lisp/byte-run.el
(byte-run--unescaped-character-literals-warning): New defun.
* src/lread.c (load_warn_unescaped_character_literals): Use new defun.
(syms_of_lread): Define symbol for new defun.
* lisp/emacs-lisp/bytecomp.el (byte-compile-from-buffer): Use new
defun.
* test/src/lread-tests.el (lread-tests--unescaped-char-literals):
test/lisp/emacs-lisp/bytecomp-tests.el
(bytecomp-tests--unescaped-char-literals): Adapt unit tests.
