The native compiler should not try to generate in rendered code
immediate floats produced by the constrain on the '=' operator.
* test/src/comp-tests.el (comp-test-78606-1): Add test.
* test/src/comp-resources/comp-test-funcs.el (comp-test-78606-1-f): New
function.
* src/comp.c (emit_mvar_rval): Check if an immediate is materializable.
* lisp/emacs-lisp/comp.el (comp-ctxt): Add 'non-materializable-objs-h'
slot.
(comp--fwprop-insn): Update call.
* lisp/emacs-lisp/comp-cstr.el (comp-cstr-=): Add parameter.
* src/frame.c (make_terminal_frame): Initialize terminal's
top_frame slot if it has not been set up yet (Bug#78966).
* src/keyboard.c (kbd_buffer_get_event): Do not assume that the
event's frame_or_window slot always produces a valid frame
(Bug#78966).
(tty_read_avail_input): Try to make sure that the input event we
create has a valid frame_or_window slot (Bug#78966). Add assertion
to that purpose.
This commit only adds the new function, the fix for the bug is
in the next commit.
* doc/lispref/parsing.texi (Using Parser): Add docs.
* lisp/treesit.el (treesit): Add shortdoc.
* src/treesit.c (treesit_get_affected_ranges): New function
extracted from treesit_call_after_change_functions.
(treesit_call_after_change_functions): Extract out.
(treesit_ensure_parsed): Return affected regions.
(Ftreesit_parser_changed_regions): New function that returns the
affected regions.
Abstract the buffering in the reader with a struct and plain functions
instead of rather unhygienic preprocessor macros.
* src/lread.c (READ_AND_BUFFER, INVALID_SYNTAX_WITH_BUFFER):
Removed, replaced with...
(readbuf_t, readbuf_grow, add_char_to_buffer, read_and_buffer)
(invalid_syntax_with_buffer): ...these new functions and struct.
(read0): Use new code instead of old preprocessor macros.
Previously, latin-1 was incorrectly assumed (bug#70988).
* src/lread.c (readchar): Convert to raw byte.
* test/src/lread-tests.el (lread-unibyte-string-source): New test.
* src/editfns.c (Ftranspose_regions): Be careful that ELisp code could
move the gap from under our feet.
* test/src/editfns-tests.el (editfns-tests--bug76124): New test.
print_error_message can throw after producing some output, so use
unwind-protect to ensure prin1-to-string-buffer is cleared.
* src/print.c (erase_prin1_to_string_buffer): New.
(Ferror_message_string): Use it to catch errors thrown in
'print_error_message'.
* test/src/print-tests.el (error-message-string-circular): Expand
test.
C99 prohibits passing zero args to macro’s ‘...’.
* lib-src/seccomp-filter.c (RULE0): New macro, which is like RULE
except with zero args. All zero-arg uses of RULE changed to RULE0.
* src/comp.c (compile_function, Fcomp__init_ctxt, syms_of_comp):
Change ‘CALLN (Fmake_hash_table)’ to ‘Fmake_hash_table (0, NULL)’,
since CALLN requires at least two args in C99.
* src/fileio.c (Finsert_file_contents): Inhibit ask-supersession
only if we're VISITing in a non-narrowed buffer (bug#78866).
* test/src/fileio-tests.el (ert--tests-dir): New var.
(fileio-tests--insert-file-contents-supersession): New test.
* src/sysdep.c (get_host_uptime) [GNU_LINUX || CYGWIN || __ANDROID__]:
New function, replacing get_host_boot_time.
Use CLOCK_BOOTTIME instead of inspecting /proc/stat.
All callers changed. This should be an even better way to fix Bug#63496.
With a Linux kernel, use /proc/stat rather than get_boot_time
to get boot time, since when Emacs is running in a container
we want the underlying host boot time, not the container boot time.
This should be a better way to fix Bug#63496.
* src/sysdep.c: Do not include boot-time.h.
(get_host_boot_time) [GNU_LINUX || CYGWIN || __ANDROID__]:
New function.
(system_process_attributes) [GNU_LINUX || CYGWIN || __ANDROID__]:
Use it instead of get_boot_time.
* src/xdisp.c (try_scrolling): Don't loop when the
partially-visible cursor line is at the top of the window. This
looping are pointless, and just makes redisplay slow for no reason.
(Bug#78766)
I noticed this issue while looking into Bug#63496.
GNU Emacs does not infer the boot time consistently on GNU/Linux
and similar platforms: android_notifications_notify_1 and
get_boot_sec use Gnulib’s boot-time module, but process-attributes
has idiosyncratic code that evidently predates that module.
The idiosyncratic code returns an unstable etime, which is a minor bug.
Simplify process-attributes by just using Gnulib.
This returns a stable etime.
It may also fix Bug#63496; if not, we can fix Gnulib later.
* src/sysdep.c: Include <boot-time.h>.
(get_up_time) [GNU_LINUX || CYGWIN || __ANDROID__]: Remove.
(system_process_attributes) [GNU_LINUX || CYGWIN || __ANDROID__]:
Use get_boot_time instead of get_up_time.
Contrary to the existing documentation here, the value of this
field cannot be the symbols open or closed. Furthermore, this
field is never set to "stop" for network connections.
* src/process.h (Lisp_Process): Improve/fix documentation of the
'status' field. (Bug#78784)
* src/profiler.c (export_log): Check if a log has been allocated first,
and return nil if it hasn't.
(Fprofiler_cpu_log, Fprofiler_memory_log): Doc fix.
* test/src/profiler-tests.el (profiler-tests-cpu-profiler)
(profiler-tests-memory-profiler): New tests.
Previously Emacs only looks for filenames like
libtree-sitter-json.so.0.0. Now Emacs also look for filenames
like libtree-sitter-json.so.15.0.
* src/treesit.c:
(treesit_load_language_push_for_each_suffix): Add versioned
candidate to candidate list too.
Specifically, use the term "display specification" more consistently to
distinguish from "display property", which is the full value of the
'display' text property.
* src/xdisp.c (find_display_property): Rename PROP to SPEC.
(Fget_display_property): Rename PROP to SPEC and improve docstring.
* lisp/emacs-lisp/subr-x.el (add-display-text-property): Rename PROP to
SPEC and improve docstring.
* doc/lispref/display.texi (Display Property): Reword documentation to
more-consistently refer to display specifications.
* java/AndroidManifest.xml.in: Update targetSdkVersion to 36.
* java/INSTALL: Document revised compilation dependencies.
* java/org/gnu/emacs/EmacsActivity.java (interceptBackGesture):
New function.
(onCreate): Invoke the same to register back gesture callbacks
on Android 16 or better.
* java/org/gnu/emacs/EmacsWindow.java (onBackInvoked): New
function.
* src/keyboard.c (lispy_function_keys): Amend with new symbols
introduced in Android API 36.
The docstring already warns against calling signal with a nil
error symbol, which is for internal use only, but we can avoid crashing
in this case.
* src/eval.c (Fsignal): Produce a "peculiar error" for more arguments
involving non-lists.
This is the NS implementation for 'display-monitor-attributes-list'.
Change implementation from IORegistry to direct introspection
of NSScreen.
* src/nsfns.m (ns_make_monitor_attribute_list): Use localizedName
selector on NSScreen. For anonymous displays, synthesize names
of the form ("%dx%d@%d,%d" width height x y).
(ns_get_name_from_ioreg) (ns_screen_name): Removed.
* src/frame.c (delete_frame): Delete initial daemon frame only
if FORCE is non-nil (Bug#78583).
(Fdelete_frame): Rewrite doc-string to mention that it can
delete an initial daemon frame if and only if FORCE is non-nil.
* etc/NEWS:
* doc/lispref/frames.texi (Deleting Frames): Mention that
'delete-frame' can delete an initial daemon frame if and only if
FORCE is non-nil.
* src/print.c (print_string_1): Renamed from 'print_string', with an
extra argument to disable nonascii escaping.
(print_string): New function.
(print_object): Use 'print_string_1', not 'strout'.
