* doc/lispref/keymaps.texi (Scanning Keymaps): Document it.
* lisp/keymap.el (make-non-key-event): New function.
* lisp/term/common-win.el (x-setup-function-keys): Mark ns events
as not being keys (bug#55940).
* src/keymap.c (Fwhere_is_internal): Filter out key sequences that
are marked as being non-keys.
* test/src/timefns-tests.el (decode-then-encode-time)
(time-equal-p-NaN-NaN, time-arith-tests):
Don’t test time functions with infinities and NaNs,
which are not Lisp time values.
* test/src/fileio-tests.el (fileio-tests--non-regular-insert): Rename
from `test-non-regular-insert` and make it into a test rather than
a broken function. Oh, and make it work while at it.
* doc/lispref/files.texi (Reading from Files): Document it.
* src/fileio.c (Finsert_file_contents): Allow specifying END for
special files (bug#18370).
d02c94090c Fix error reporting in process-async-https-with-delay
9a4862a973 * doc/misc/org.org: Remove spurious markup.
768ed1476a Make Tramp version check more robust
7f778c6943 Fix debugging with GDB when a breakpoint has multiple loca...
25e53e9391 ; * lisp/files.el (file-expand-wildcards): Doc fix.
3ea9357d10 Update documentation of 'aset' and 'store-substring'
# Conflicts:
# lisp/files.el
* test/src/process-tests.el (process-async-https-with-delay): Use
'plist-get' instead of 'assq' in testing for a connection error.
The 'status' variable is a plist, not an alist. (Bug#55849)
As suggested by Stefan Monnier.
* src/lread.c (read_escape):
Signal an error for ?\LF since it cannot reasonably be intended.
* test/src/lread-tests.el (lread-escaped-lf): Update test.
* etc/NEWS: Announce.
The old -1 value was an artefact of the reader implementation.
* src/lread.c (read_escape): Remove the `stringp` argument; assume
character literal syntax. Never return -1.
(read_string_literal): Handle string-specific escape semantics here
and simplify.
* test/src/lread-tests.el (lread-escaped-lf): New test.
Restructure the reader to be nonrecursive so that it is not limited by
the C stack or crashes Emacs when reading deeply nested data.
This also improves performance.
A few minor bugs were fixed:
- (a .{NBSP}b) where {NBSP} is a non-breaking space (U+00A0) is now
the dotted pair (a . b), not the 3-element list (a \. b), since U+00A0
is treated as whitespace everywhere else.
- #_ with no symbol following is now equivalent to ## (empty interned
symbol), not #: (empty uninterned symbol).
* src/alloc.c (garbage_collect): Call mark_lread.
* src/lread.c (readevalloop): Use read0 instead of read_list.
(stackbufsize): Increase to 1024, now that read0 isn't recursive.
(invalid_radix_integer): Buffer overflow check.
(read1, read_list, read_vector): Remove.
(read_char_literal, read_string_literal)
(hash_table_from_plist, record_from_list, vector_from_rev_list)
(bytecode_from_rev_list, char_table_from_rev_list)
(sub_char_table_from_rev_list, string_props_from_rev_list)
(read_bool_vector, skip_lazy_string, symbol_char_span)
(skip_space_and_comments)
(enum read_entry_type, struct read_stack_entry, struct read_stack)
(rdstack, mark_lread, read_stack_top, read_stack_pop)
(read_stack_empty_p, grow_read_stack, read_stack_push): New.
(read0): Rewrite to be nonrecursive.
* test/src/lread-tests.el (lread-deeply-nested, lread-misc): New tests.
* src/atimer.c (set_alarm): If the atimer has already expired, signal
it right away instead of postponing it further. Previously this could
occur repeatedly, blocking atimers indefinitely.
Also only use `alarm` as fallback if `setitimer` is unavailable, not
both at the same time (which makes no sense, and they both typically
use the same mechanism behind the curtains).
* test/src/eval-tests.el (eval-tests/funcall-with-delayed-message):
New test, verifying proper functioning of funcall-with-delayed-message
which also serves as test for this bug (which also caused
debug-timer-check to fail, but that test is only run when Emacs is
built with enable-checking).
When printing a circular list and `print-circle` is nil, use a
somewhat more meaningful ". #N" tail index. The previous method for
calculating that index was based on Floyd circularity detection being
used so it had been broken ever since the change to Brent's algorithm.
The new index is correct with respect to the start of the list itself
which is what it used to be before being completely broken.
It does not take into account the nesting depth of the list context.
* src/print.c (struct print_stack_entry, print_object):
Keep track of the tortoise index (which is cheap) instead of trying
to derive it from the printed element index.
* test/src/print-tests.el (print-test-rho, print-circular):
New test.
Introduce explicit stacks for traversing common data types during
printing: conses, vectors, records, byte-code, hash-tables and
char-tables, all previously traversed using recursion in C. This
greatly reduces the risk of crashing Emacs from C stack overflow
when printing deeply nested data.
* src/print.c (Fprinc, print, PRINT_CIRCLE_CANDIDATE_P):
Special-case Fprinc with a plain string argument to eliminate the need
for keeping track of print_depth during the preprocessing phase.
This also improves performance.
(struct print_pp_entry, struct print_pp_stack, ppstack)
(grow_pp_stack, pp_stack_push_value, pp_stack_push_values)
(pp_stack_empty_p, pp_stack_pop):
New stack for preprocessing.
(print_preprocess):
Make mostly nonrecursive, except for string properties.
(enum print_entry_type, struct print_stack_entry)
(struct print_stack, prstack, grow_print_stack)
(print_stack_push, print_stack_push_vector):
New stack for printing.
(print_vectorlike, print_object):
Make mostly nonrecursive, except for string properties and some less
heavily used types.
* test/src/print-tests.el (print-deeply-nested):
New test.
* doc/lispref/buffers.texi (Buffer Modification): Adjust
documentation.
* src/buffer.c (Frestore_buffer_modified_p): Fix up the logic
around `autosaved': It means "the buffer is modified, and also
autosaved".
* doc/lispref/buffers.texi (Buffer Modification): Document
buffer-modified-p returning `autosaved'.
* lisp/subr.el (with-silent-modifications): Use
restore-buffer-modified-p instead of altering the buffer modiff
(since this has other side effects like not updating after async
`display' changes.
* src/buffer.c (Fbuffer_modified_p): Allow returning whether the
buffer has been autosaved after changes.
(Frestore_buffer_modified_p): Allow adjusting whether the buffer
has been autosaved after changes.
* src/fileio.c (Fdo_auto_save): Refill the doc string.
* src/timefns.c (Fencode_time): Add support for a 6-elt list arg.
Requested by Max Nikulin for Org (bug#54764).
* test/src/timefns-tests.el (encode-time-alternate-apis): New test.
Since this function is commonly used as a sorting predicate
where it is time-critical, this is a useful optimisation.
* src/fns.c (Fstring_lessp): Add fast path for the common case
when both arguments are unibyte.
* test/src/fns-tests.el (fns-tests--string-lessp-cases)
(fns-tests-string-lessp): New test.
Make sure that the value added to the `read_objects_completed` set is
the one we actually return; previously this wasn't the case for conses
because of an optimisation (bug#54501).
Also add a check for vacuous self-references such as #1=#1# instead of
returning a nonsense value from thin air.
* src/lread.c (read1): Treat numbered conses correctly as described
above. Detect vacuous self-references.
* test/src/lread-tests.el (lread-test-read-and-print)
(lread-test-circle-cases, lread-circle): Add tests.
mod-test.c should not use Gnulib code, as that creates
unnecessary coupling between Emacs internals and this test module.
Also Gnulib code is compiled without -fPIC whereas mod-test.c
needs -fPIC and recompiling Gnulib with -fPIC would be too painful.
* src/Makefile.in (LIB_NANOSLEEP): New macro.
(LIBES): Use it.
* test/Makefile.in (REPLACE_FREE, FREE_SOURCE_0, FREE_SOURCE_0):
Remove. All uses removed.
(LIB_CLOCK_GETTIME, LIB_NANOSLEEP): New macros.
(MODULE_CFLAGS): Do not -I from lib as that would include Gnulib
modifications to standard .h files (e.g., "#define nanosleep
rpl_nanosleep") and we don’t want the Gnulib replacements.
Instead, for gmp.h (on platforms lacking <gmp.h>) simply '-I.'
with a suitable gmp.h.
(gmp.h): New rule to create a suitable gmp.h.
($(test_module)): Depend on config.h since the code uses config.h.
Depend on gmp.h if we need to create a suitable one.
If compiling mini-gmp.h, compile the original one and not
the Emacs-specific one, to lessen coupling with Emacs internals.
Link with LIB_CLOCK_GETTIME and LIB_NANOSLEEP.
(clean): Remove gmp.h.
* test/src/emacs-module-resources/mod-test.c: Don’t include timespec.h.
All uses of timespec.h APIs changed to use system-supplied APIs.
Change _Static_assert to plain assert, so that we needn’t rely
on Gnulib’s _Static_assert.
(timespec_le) [CLOCK_REALTIME]: New function.
Change use of timespec_cmp changed to use this instead.
(Fmod_test_sleep_until, Fmod_test_nanoseconds):
Define these functions and their Lisp names mod-test-sleep-until
and mod-test-nanoseconds only if CLOCK_REALTIME,
since they now won’t work on platforms lacking CLOCK_REALTIME.
(Fmod_test_nanoseconds): Just use _Static_assert since it
should work on all platforms.
* test/src/emacs-module-tests.el (mod-test-sleep-until)
(mod-test-nanoseconds, mod-test-double):
Skip test if the corresponding functione is not defined.
`color-values-from-color-spec` (new in Emacs 28) erroneously accepted
leading whitespace and hex floats in rgbi: components.
Reported by Philip Kaludercic.
* src/xfaces.c (parse_float_color_comp): Disallow leading whitespace
and hex floats.
* test/src/xfaces-tests.el
(xfaces-internal-color-values-from-color-spec): Add test cases.
* test/src/doc-tests.el
(doc-tests-documentation/autoloaded-macro): Adjust test -- rx is
loaded in nativecomp Emacsen when the tests are run, apparently.
* test/src/emacs-module-tests.el (module/describe-function-1):
Prevent string mismatch caused by line breaks after filling;
the macOS dynamic library suffix (.dylib) is longer than on other
platforms.
Make the context switch code handle buffer-local variables more
correctly by reusing the code originally written for `backtrace-eval`.
This has the side benefit of making the `saved_value` field unused.
* src/lisp.h (enum specbind_tag): Remove `saved_value` field.
(rebind_for_thread_switch, unbind_for_thread_switch): Delete decls.
(specpdl_unrewind): Declare function.
* src/eval.c (specpdl_saved_value): Delete function.
(specbind): Delete the code related to `saved_value`, and consolidate
common code between the different branches.
(rebind_for_thread_switch, -unbind_for_thread_switch): Move to `thread.c`.
(specpdl_unrewind): New function, extracted from `backtrace_eval_unrewind`.
Use `SET_INTERNAL_THREAD_SWITCH`. Skip the buffer & excursion unwinds
depending on new arg `vars_only`.
(backtrace_eval_unrewind): Use it.
(mark_specpdl): Don't mark `saved_value`.
* src/thread.c (rebind_for_thread_switch, unbind_for_thread_switch):
Move from `eval.c` and rewrite using `specpdl_unrewind`.
* test/src/thread-tests.el (threads-test-bug48990): New test.
* test/Makefile.in (test_template): Add a + as suggested by make:
"warning: jobserver unavailable: using -j1. Add '+' to parent make rule".
* lisp/emacs-lisp/comp-cstr.el (comp-cstr-set-range-for-arithm):
When one of the two sources is negated revert to set dst as
number.
* test/src/comp-tests.el (comp-tests-type-spec-tests): Add test to
verify this is effective.
* test/lisp/electric-tests.el (electric-pair-define-test-form):
* test/src/buffer-tests.el (buffer-tests--make-test-name): Silence
byte-compiler by defining defuns also at run time.
* test/src/filelock-tests.el (filelock-tests--fixture): Make it a
defmacro. Adapt callees.
(filelock-tests-unlock-spoiled, filelock-tests-kill-buffer-spoiled):
Simplify.
(filelock-tests-detect-external-change): New test
* doc/lispref/display.texi (Logging Messages): Document it.
* src/xdisp.c (message_dolog): Add sanity checking.
(syms_of_xdisp): Make Vmessages_buffer_name into a defvar
(bug#27170).
* doc/lispref/streams.texi (Output Variables): Document it.
* src/print.c (print_vectorlike): Use the variable.
(syms_of_print): New variable print-unreadable-function
(bug#52566).
Fchar_syntax did not convert unibyte characters to multibyte when the
current buffer was unibyte, in contrast to `char-syntax` in
byte-compiled code (bug#53260).
* src/bytecode.c (exec_byte_code): Call out to Fchar_syntax;
the dynamic frequency is too low to justify inlining here, and it
did lead to implementations diverging.
* src/syntax.c (Fchar_syntax): Convert non-ASCII unibyte values to
multibyte.
* test/src/syntax-tests.el (syntax-char-syntax): New test.
