GitLab

* window.c (Fset_window_configuration): Remove handling of
auto-buffer-name window parameter.

* buffer.c (Fset_buffer_major_mode): Signal an error
if the buffer is dead.
(Fother_buffer, other_buffer_safely): Remove redundant
nested declaration.

* buffer.h (record_unwind_current_buffer): New function.
* bytecode.c, dispnew.c, editfns.c, fileio.c, fns.c, insdel.c:
* keyboard.c, keymap.c, minibuf.c, print.c, process.c, textprop.c:
* undo.c, window.c: Adjust users.
* buffer.c (set_buffer_if_live): Fix comment.

* buffer.h (set_buffer_internal): Make it BUFFER_INLINE.
* buffer.c (set_buffer_if_live): Use set_buffer_internal.
* coding.c (decode_coding): Omit redundant test.
* fileio.c (decide_coding_unwind): Likewise.
* fns.c (secure_hash): Likewise.
* insdel.c (modify_region): Likewise.
* keyboard.c (command_loop_1): Likewise.
* print.c (PRINTFINISH): Likewise.
* xdisp.c (run_window_scroll_functions): Use set_buffer_internal.

(frame_garbaged, display_completed, delayed_size_change)
(fonts_changed_p, add_window_display_history)
(add_frame_display_history, verify_row_hash)
(adjust_glyph_matrix, clear_window_matrices, glyph_row_slice_p)
(row_equal_p, realloc_glyph_pool)
(allocate_matrices_for_frame_redisplay)
(showing_window_margins_p)
(adjust_frame_glyphs_for_frame_redisplay)
(build_frame_matrix_from_leaf_window, make_current)
(mirrored_line_dance, mirror_line_dance, update_frame)
(update_window_tree, update_single_window)
(check_current_matrix_flags, update_window, update_text_area)
(update_window_line, set_window_update_flags, scrolling_window)
(update_frame_1, scrolling, buffer_posn_from_coords)
(do_pending_window_change, change_frame_size)
(change_frame_size_1, sit_for):
Use bool for boolean.
(clear_glyph_matrix_rows): Rename from enable_glyph_matrix_rows,
and remove last int (actually boolean) argument, which was always 0.
All callers changed.
* dispextern.h, frame.h, lisp.h: Reflect above API changes.
* dispextern.h (struct composition_it): Use bool for boolean.
(struct glyph_matrix): Don't assume buffer sizes can fit in 'int'.
(struct bidi_it): Use unsigned:1, not int, for boolean prev_was_pdf.
* dired.c (file_name_completion):
Use bool for boolean.  (This was missed in an earlier change.)

* window.c (Fset_window_configuration): Revert first part of
last change.

dialog_return.
(windowShouldClose:): Use stop instead of stopModalWithCode.
(clicked:): Ditto, and also set dialog_return.
(timeout_handler:): Use stop instead of abortModal.  Send a dummy
event.
(runDialogAt:): Make ret Lisp_Object.  Set it from dialog_return when
modal loop returns.

* nsterm.h (NSPanel): New class variable dialog_return.

Fixes: debbugs:12258

* composite.c (find_composition, composition_gstring_p)
(composition_reseat_it, find_automatic_composition):
* data.c (let_shadows_buffer_binding_p)
(let_shadows_global_binding_p, set_internal, make_blv)
(Fmake_variable_buffer_local, Fmake_local_variable)
(Fmake_variable_frame_local, arithcompare, cons_to_unsigned)
(cons_to_signed, arith_driver):
* dbusbind.c (xd_in_read_queued_messages):
* dired.c (directory_files_internal, file_name_completion):
Use bool for booleans.
* dired.c (file_name_completion):
* process.h (fd_callback):
Omit int (actually boolean) argument.  It wasn't being used.
All uses changed.
* composite.h, lisp.h: Reflect above API changes.

* cmds.c (move_point, Fself_insert_command):
* coding.h (struct composition status, struct coding_system):
* coding.c (detect_coding_utf_8, encode_coding_utf_8)
(detect_coding_utf_16, encode_coding_utf_16, detect_coding_emacs_mule)
(emacs_mule_char, decode_coding_emacs_mule)
(encode_coding_emacs_mule, detect_coding_iso_2022)
(decode_coding_iso_2022, encode_invocation_designation)
(encode_designation_at_bol, encode_coding_iso_2022)
(detect_coding_sjis, detect_coding_big5, decode_coding_sjis)
(decode_coding_big5, encode_coding_sjis, encode_coding_big5)
(detect_coding_ccl, encode_coding_ccl, decode_coding_raw_text)
(encode_coding_raw_text, detect_coding_charset)
(decode_coding_charset, encode_coding_charset, detect_eol)
(detect_coding, get_translation_table, produce_chars)
(consume_chars, reused_workbuf_in_use)
(make_conversion_work_buffer, code_conversion_save)
(decode_coding_object, encode_coding_object)
(detect_coding_system, char_encodable_p)
(Funencodable_char_position, code_convert_region)
(code_convert_string, code_convert_string_norecord)
(Fset_coding_system_priority):
* fileio.c (Finsert_file_contents):
Use bool for booleans.
* coding.h, lisp.h: Reflect above API changes.
* coding.c: Remove unnecessary static function decls.
(detect_coding): Use unsigned, not signed, to copy an unsigned field.
(decode_coding, encode_coding, decode_coding_gap): Return 'void',
not a boolean 'int', since callers never look at the return value.
(ALLOC_CONVERSION_WORK_AREA): Assume caller returns 'void', not 'int'.
* coding.h (decoding_buffer_size, encoding_buffer_size)
(emacs_mule_string_char): Remove unused extern decls.
(struct iso_2022_spec, struct coding_system):
Use 'unsigned int : 1' for boolean fields, since there's more than one.
(struct emacs_mule_spec): Remove unused field 'full_support'.
All initializations removed.
* cmds.c (internal_self_insert): Don't assume EMACS_INT fits in 'int'.

* alloc.c (mark_maybe_pointer): Handle MEM_TYPE_SPARE.
(valid_lisp_object_p): Likewise.

* window.c (Fset_window_configuration): Record any window's old
buffer if it's replaced (see Bug#8789).  If the new current
buffer doesn't appear in the selected window, go to its old
point (Bug#12208).

* alloc.c (enum mem_type): New memory type.
(refill_memory_reserve): Use new type for spare memory.
This prevents live_cons_p and live_string_p from incorrect
detection of uninitialized objects from spare memory as live.

* Makefile.in (.PHONY): versioclean -> versionclean.

* data.c (Qcons, Qfloat, Qmisc, Qstring, Qvector):
* window.c (Qwindow_valid_p, decode_valid_window):
Now static, not extern.
* data.c (Qinterval): Remove; unused.
(syms_of_data): Do not define 'interval'.
* lisp.h (Qinteger, Qstring, Qmisc, Qvector, Qfloat, Qcons):
* window.h (decode_valid_window):
Remove decls.

* character.c (lisp_string_width, string_count_byte8)
(string_escape_byte8):
* charset.c (charset_map_loaded, load_charset_map, read_hex):
(load_charset_map_from_file, map_charset_chars)
(Fdefine_charset_internal, define_charset_internal)
(Fdeclare_equiv_charset, find_charsets_in_text)
(Ffind_charset_region, char_charset, Fiso_charset):
* chartab.c (sub_char_table_ref, sub_char_table_ref_and_range)
(sub_char_table_set, sub_char_table_set_range)
(char_table_set_range, optimize_sub_char_table)
(map_sub_char_table):
Use bool for boolean.
* character.c (str_to_unibyte): Omit last boolean argument; it was
always 0.  All callers changed.
* character.h, charset.h: Adjust to match previous changes.
* character.h (char_printable_p): Remove decl of nonexistent function.
* charset.h (struct charset): Members code_linear_p, iso_chars_96,
ascii_compatible_p, supplementary_p, compact_codes_p, unified_p
are all boolean, so make them single-bit bitfields.

It was meant to be temporary and it often doesn't work,
because when IDX has side effects the behavior of IDX==IDX
is undefined.  See Stefan Monnier in
<http://lists.gnu.org/archive/html/emacs-devel/2012-08/msg00762.html>.

(FUNCTIONP): Use it.
* src/eval.c (Ffunctionp): Use it.

as that's faster and simpler than static storage.  Don't bother
with the g_main_context_query overhead if g_main_context_pending
says no events are pending.
(gfds, gfds_size): Remove these static vars.
(xgselect_initialize): Remove; no longer needed.
All uses and decls removed.

All uses removed.  This leftover from old code was always 0.

* casefiddle.c (casify_object, casify_region):
* casetab.c (set_case_table):
* category.c, category.h (word_boundary_p):
* category.h (CHAR_HAS_CATEGORY):
Use bool for booleans, instead of int.

MS-Windows adaptations for 2012-08-25T04:04:08Z!eggert@cs.ucla.edu and 2012-08-25T04:27:32Z!eggert@cs.ucla.edu.

 lib/makefile.w32-in ($(BLD)/execinfo.$(O), execinfo.h): New targets.
 (GNULIBOBJS): Add $(BLD)/execinfo.$(O).
 src/makefile.w32-in ($(BLD)/alloc.$(O)): Depend on $(GNU_LIB)/execinfo.h.

Merge from gnulib, incorporating:
2012-08-24 execinfo: port to FreeBSD
2012-08-22 execinfo: new module
* admin/merge-gnulib (GNULIB_MODULES): Add execinfo.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib/execinfo.c, lib/execinfo.in.h, m4/execinfo.m4: New files.
* src/alloc.c [ENABLE_CHECKING]: Include <execinfo.h>.
(die) [ENABLE_CHECKING]: Print a backtrace if available.
* src/Makefile.in (LIB_EXECINFO): New macro.
(LIBES): Use it.

* bytecode.c (exec_byte_code):
* callint.c (check_mark, Fcall_interactively):
* callproc.c (Fcall_process, add_env, child_setup, getenv_internal_1)
(getenv_internal, sync_process_alive, call_process_exited):
* lisp.h (USE_SAFE_ALLOCA):
Use bool for booleans, instead of int.
* lisp.h, process.h: Adjust prototypes to match above changes.
* callint.c (Fcall_interactively): Don't assume the mark's
offset fits in 'int'.

* buffer.c (reset_buffer_local_variables)
(buffer_lisp_local_variables, Fset_buffer_modified_p)
(Frestore_buffer_modified_p, Fset_buffer_multibyte):
(overlays_at, overlays_in, mouse_face_overlay_overlaps)
(overlay_touches_p, overlay_strings, Foverlay_put)
(report_overlay_modification, call_overlay_mod_hooks):
(mmap_enlarge, mmap_set_vars):
* buffer.h (buffer_has_overlays, uppercasep, lowercasep):
Use bool for booleans, instead of int.
* buffer.c (compact_buffer, mmap_free_1): Return void, not int,
since the 1-or-0 return value is always ignored anyway.
(mmap_initialized_p):
* buffer.h (struct buffer_text.inhibit_shrinking): Now bool, not int.
* buffer.h, lisp.h: Adjust prototypes to match above changes.

This is a bit more readable, and makes the text segment of bidi.o
0.4% smaller on my platform (GCC 4.7.1 x86-64, Fedora 15).
Presumably it's faster too.
(bidi_initialized, bidi_ignore_explicit_marks_for_paragraph_level):
Now bool.
(bidi_cache_find_level_change, bidi_cache_iterator_state)
(bidi_unshelve_cache, bidi_init_it, bidi_count_bytes)
(bidi_char_at_pos, bidi_fetch_char, bidi_paragraph_init)
(bidi_explicit_dir_char, bidi_level_of_next_char)
(bidi_find_other_level_edge, bidi_move_to_visually_next):
Use bool for booleans, instead of int.
* dispextern.h (bidi_init_it, bidi_paragraph_init)
(bidi_unshelve_cache): Adjust decls to match code.

* keyboard.c (Fposn_at_x_y): Do not allow internal window as
argument.

* atimer.h: Include <stdbool.h>.

compile time tests instead of run time tests on systems that do
not use them.
(FRAME_MAC_P): Remove leftover from deleted code.
* frame.c (syms_of_frame): Remove leftover from deleted code.

Otherwise, the compiler complains about (A?B:C) where B is void
and C is Lisp_Object.  This fixes an incompatibility with Sun C	5.12.
(fontset_add): Return void, for FONTSET_ADD.

(gc_in_progress, abort_on_gc)
(setjmp_tested_p) [!GC_SAVE_REGISTERS_ON_STACK && !GC_SETJMP_WORKS]:
(dont_register_blocks) [GC_MALLOC_CHECK]:
(suppress_checking) [ENABLE_CHECKING]: Now bool, not int.
(check_string_bytes, make_specified_string, memory_full)
(live_string_p, live_cons_p, live_symbol_p, live_float_p)
(live_misc_p, live_vector_p, live_buffer_p, mark_maybe_object)
(mark_stack, valid_pointer_p, make_pure_string)
(Fgarbage_collect, survives_gc_p, gc_sweep):
Use bool for booleans, instead of int.
(test_setjmp) [!GC_SAVE_REGISTERS_ON_STACK && !GC_SETJMP_WORKS]:
Remove unused local.
* alloc.c (PURE_POINTER_P):
* lisp.h (STRING_MULTIBYTE): Document that it returns a boolean.
* editfns.c (Fformat):
* fileio.c (Fexpand_file_name, Fsubstitute_in_file_name)
(Fdo_auto_save):
* fns.c (sweep_weak_table):
* lisp.h (suppress_checking, push_message, survives_gc_p)
(make_pure_string, gc_in_progress, abort_on_gc):
* lread.c (readchar, read1):
* print.c (Fprin1_to_string):
* xdisp.c (push_message):
Use bool for booleans affected directly or indirectly by
alloc.c's changes.

* fontset.c (set_fontset_id, set_fontset_name, set_fontset_ascii)
(set_fontset_base, set_fontset_frame, set_fontset_nofont_face)
(set_fontset_default, set_fontset_fallback): Rename from their
upper-case counterparts, and make them functions rather than macros.
This is more consistent with the other recently-introduced setters.
These don't need to be inline, since they're local.

the loop.

Fixes: debbugs:12247

This fixes a performance regression introduced by the recent
addition of vcopy.  This means 'vcopy' will need to be modified
for a copying collector, but that's OK.  Also, tighten the
checking in the assertion.

 src/w32uniscribe.c (uniscribe_shape): Fix producing gstring
 components for RTL text.  Adjust X-OFFSET of each non-base glyph
 for the width of the base character, according to what
 x_draw_composite_glyph_string_foreground expects.  Generate
 WADJUST value according to composition_gstring_width's
 expectations, to produce correct width of the composed character.
 Reverse the sign of the DU offset produced by ScriptPlace.

* lisp.h (vcopy): New function to copy data into vector.
* dispnew.c (Fframe_or_buffer_changed_p): Use AREF and ASET.
* fns.c (Ffillarray): Use ASET.
* keyboard.c (timer_check_2): Use AREF and ASET.
(append_tool_bar_item, Frecent_keys): Use vcopy.
* lread.c (read_vector): Use ASET.
* msdos.c (Frecent_doskeys): Use vcopy.
* xface.c (Finternal_copy_lisp_face): Use vcopy.
(Finternal_merge_in_global_face): Use ASET and vcopy.
* xfont.c (xfont_list_pattern): Likewise.

* window.c (Fwindow_point): For the selected window always return
the position of its buffer's point.
(Fset_window_point): For the selected window always go in its
buffer to the specified position.
* window.el (window-point-1, set-window-point-1): Remove.
(window-in-direction, record-window-buffer)
(set-window-buffer-start-and-point, split-window-below)
(window--state-get-1, display-buffer-record-window): Replace
calls to window-point-1 and set-window-point-1 by calls to
window-point and set-window-point respectively.

* fontset.c (SET_FONTSET_ID, SET_FONTSET_NAME, SET_FONTSET_ASCII)
(SET_FONTSET_BASE, SET_FONTSET_FRAME, SET_FONTSET_NOFONT_FACE)
(SET_FONTSET_DEFAULT, SET_FONTSET_FALLBACK): New macros.
Adjust users.

Presumably it does work fine on all platforms currently building
Emacs, but the autoconf manual says not to rely on it.