Github/emacs
2
0
Fork 1
mirror of git://git.sv.gnu.org/emacs.git synced 2026-02-17 10:27:41 +00:00
Code Issues Projects Releases Packages Wiki Activity
emacs/src/scroll.c
Paul Eggert d311d28c3f * alloc.c (pure_bytes_used_lisp, pure_bytes_used_non_lisp): 
(allocate_vectorlike, buffer_memory_full, struct sdata, SDATA_SIZE)
(string_bytes, check_sblock, allocate_string_data):
(compact_small_strings, Fmake_bool_vector, make_string)
(make_unibyte_string, make_multibyte_string)
(make_string_from_bytes, make_specified_string)
(allocate_vectorlike, Fmake_vector, find_string_data_in_pure)
(make_pure_string, make_pure_c_string, make_pure_vector, Fpurecopy)
(mark_vectorlike):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(allocate_pseudovector):
Use int, not EMACS_INT, where int is wide enough.
(inhibit_garbage_collection, Fgarbage_collect):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* bidi.c (bidi_mirror_char): Use EMACS_INT, not int, where
int might not be wide enough.
(bidi_cache_search, bidi_cache_find, bidi_init_it)
(bidi_count_bytes, bidi_char_at_pos, bidi_fetch_char)
(bidi_at_paragraph_end, bidi_find_paragraph_start)
(bidi_paragraph_init, bidi_resolve_explicit, bidi_resolve_weak)
(bidi_level_of_next_char, bidi_move_to_visually_next):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* buffer.c (copy_overlays, Fgenerate_new_buffer_name)
(Fkill_buffer, Fset_buffer_major_mode)
(advance_to_char_boundary, Fbuffer_swap_text)
(Fset_buffer_multibyte, overlays_at, overlays_in)
(overlay_touches_p, struct sortvec, record_overlay_string)
(overlay_strings, recenter_overlay_lists)
(adjust_overlays_for_insert, adjust_overlays_for_delete)
(fix_start_end_in_overlays, fix_overlays_before, modify_overlay)
(Fmove_overlay, Fnext_overlay_change, Fprevious_overlay_change)
(Foverlay_recenter, last_overlay_modification_hooks_used)
(report_overlay_modification, evaporate_overlays, enlarge_buffer_text):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(validate_region): Omit unnecessary test for b <= e, since
that's guaranteed by the previous test.
(adjust_overlays_for_delete): Avoid pos + length overflow.
(Fmove_overlay, Fdelete_overlay, add_overlay_mod_hooklist)
(report_overlay_modification):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Foverlays_at, Fnext_overlay_change, Fprevious_overlay_change):
Omit pointer cast, which isn't needed anyway, and doesn't work
after the EMACS_INT -> ptrdiff_t change.
* buffer.h: Adjust decls to match defn changes elsewhere.
(struct buffer_text, struct buffer):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
Use EMACS_INT, not int, where int might not be wide enough.
* bytecode.c (exec_byte_code): Use ptrdiff_t, not int, to avoid
needless 32-bit limit on 64-bit hosts.  Remove unnecessary
memory-full test.  Use EMACS_INT, not ptrdiff_t or int, where
ptrdiff_t or int might not be wide enough.
* callint.c (Fcall_interactively):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* callproc.c (call_process_kill, Fcall_process):
Don't assume pid_t fits into an Emacs fixnum.
(call_process_cleanup, Fcall_process, child_setup):
Don't assume pid_t fits into int.
(call_process_cleanup, Fcall_process, delete_temp_file)
(Fcall_process_region):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Fcall_process): Simplify handling of volatile integers.
Use int, not EMACS_INT, where int will do.
* casefiddle.c (casify_object, casify_region, operate_on_word)
(Fupcase_word, Fdowncase_word, Fcapitalize_word):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(casify_object): Avoid integer overflow when overallocating buffer.
* casetab.c (set_identity, shuffle): Prefer int to unsigned when
either works.
* category.c (Fchar_category_set): Don't assume fixnum fits in int.
* category.h (CATEGORYP): Don't assume arg is nonnegative.
* ccl.c (GET_CCL_INT): Remove; no longer needed, since the
integers are now checked earlier.  All uses replaced with XINT.
(ccl_driver):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
For CCL_MapSingle, check that content and value are in int range.
(resolve_symbol_ccl_program): Check that vector header is in range.
Always copy the vector, so that we can check its contents reliably
now rather than having to recheck each instruction as it's being
executed.  Check that vector words fit in 'int'.
(ccl_get_compiled_code, Fregister_ccl_program)
(Fregister_code_conversion_map): Use ptrdiff_t, not int, for
program indexes, to avoid needless 32-bit limit on 64-bit hosts.
(Fccl_execute, Fccl_execute_on_string): Check that initial reg
contents are in range.
(Fccl_execute_on_string): Check that status is in range.
* ccl.h (struct ccl_program.idx): Now ptrdiff_t, not int.
* character.c (char_resolve_modifier_mask, Fchar_resolve_modifiers):
Accept and return EMACS_INT, not int, because callers can pass values
out of 'int' range.
(c_string_width, strwidth, lisp_string_width, chars_in_text)
(multibyte_chars_in_text, parse_str_as_multibyte)
(str_as_multibyte, count_size_as_multibyte, str_to_multibyte)
(str_as_unibyte, str_to_unibyte, string_count_byte8)
(string_escape_byte8, Fget_byte):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Funibyte_string): Use CHECK_CHARACTER, not CHECK_NATNUM, to
avoid mishandling large integers.
* character.h: Adjust decls to match defn changes elsewhere.
* charset.c (load_charset_map_from_file, find_charsets_in_text)
(Ffind_charset_region):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(load_charset_map_from_file): Redo idx calculation to avoid overflow.
(load_charset_map_from_vector, Fdefine_charset_internal):
Don't assume fixnum fits in int or unsigned int.
(load_charset_map_from_vector, Fmap_charset_chars):
Remove now-unnecessary CHECK_NATNUMs.
(Fdefine_charset_internal): Check ranges here, more carefully.
* chartab.c (Fmake_char_table, Fset_char_table_range)
(uniprop_get_decoder, uniprop_get_encoder):
Don't assume fixnum fits in int.
* cmds.c (move_point): New function, that does the gist of
Fforward_char and Fbackward_char, but does so while checking
for integer overflow more accurately.
(Fforward_char, Fbackward_char, internal_self_insert): Use it.
(Fforward_line, Fend_of_line, internal_self_insert)
(internal_self_insert):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
Fix a FIXME, by checking for integer overflow when calculating
target_clm and actual_clm.
* coding.c (detect_coding_XXX, encode_coding_XXX, CODING_DECODE_CHAR)
(ASSURE_DESTINATION, coding_alloc_by_realloc)
(coding_alloc_by_making_gap, alloc_destination)
(detect_coding_utf_8, encode_coding_utf_8, decode_coding_utf_16)
(encode_coding_utf_16, detect_coding_emacs_mule)
(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, decode_coding_ccl)
(encode_coding_ccl, encode_coding_raw_text)
(detect_coding_charset, decode_coding_charset)
(encode_coding_charset, detect_eol, decode_eol, produce_chars)
(produce_composition, produce_charset, produce_annotation)
(decode_coding, handle_composition_annotation)
(handle_charset_annotation, consume_chars, decode_coding_gap)
(decode_coding_object, encode_coding_object, detect_coding_system)
(Ffind_coding_systems_region_internal, Fcheck_coding_systems_region)
(code_convert_region, code_convert_string)
(Fdefine_coding_system_internal):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(setup_iso_safe_charsets, consume_chars, Funencodable_char_position)
(Fdefine_coding_system_internal):
Don't assume fixnums fit in int.
(decode_coding_gap, decode_coding_object, encode_coding_object)
(Fread_coding_system, Fdetect_coding_region, Funencodable_char_position)
(Fcheck_coding_systems_region):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Ffind_operation_coding_system): NATNUMP can eval its arg twice.
(Fdefine_coding_system_internal): Check for charset-id overflow.
* coding.h: Adjust decls to match defn changes elsewhere.
(struct coding_system):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* composite.c (get_composition_id, find_composition)
(run_composition_function, update_compositions)
(compose_text, composition_gstring_put_cache)
(composition_gstring_p, composition_gstring_width)
(fill_gstring_header, fill_gstring_body, autocmp_chars)
(composition_compute_stop_pos, composition_reseat_it)
(composition_update_it, struct position_record)
(find_automatic_composition, composition_adjust_point)
(Fcomposition_get_gstring, Ffind_composition_internal):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(update_compositions):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* composite.h: Adjust decls to match defn changes elsewhere.
(struct composition):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* data.c (let_shadows_buffer_binding_p, let_shadows_global_binding_p):
Do not attempt to compute the address of the object just before a
buffer; this is not portable.
(Faref, Faset):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Faset): Use int, not EMACS_INT, where int is wide enough.
(Fstring_to_number): Don't assume fixnums fit in int.
(Frem): Don't assume arg is nonnegative.
* dbusbind.c (xd_append_arg): Check for integers out of range.
(Fdbus_call_method): Don't overflow the timeout int.
* dired.c (directory_files_internal, file_name_completion, scmp)
(file_name_completion_stat):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(file_name_completion): Don't overflow matchcount.
(file_name_completion_stat): Use SAFE_ALLOCA, not alloca.
* dispextern.h: Adjust decls to match defn changes elsewhere.
(struct text_pos, struct glyph, struct bidi_saved_info)
(struct bidi_string_data, struct bidi_it, struct composition_it)
(struct it):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(struct display_pos, struct composition_it, struct it):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* dispnew.c (increment_matrix_positions)
(increment_row_positions, mode_line_string)
(marginal_area_string):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(change_frame_size_1, Fredisplay):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(duration_to_sec_usec): New function, to check for overflow better.
(Fsleep_for, sit_for): Use it.
* doc.c (get_doc_string, store_function_docstring):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(get_doc_string, Fsnarf_documentation):
Use int, not EMACS_INT, where int is wide enough.
(get_doc_string):
Use SAFE_ALLOCA, not alloca.
Check for overflow when converting EMACS_INT to off_t.
* doprnt.c (doprnt):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* editfns.c (init_editfns, Fuser_uid, Fuser_real_uid):
Don't assume uid_t fits into fixnum.
(buildmark, Fgoto_char, overlays_around, find_field, Fdelete_field)
(Ffield_string, Ffield_string_no_properties, Ffield_beginning)
(Ffield_end, Fconstrain_to_field, Fline_beginning_position)
(Fline_end_position, Fprevious_char, Fchar_after, Fchar_before)
(general_insert_function)
(Finsert_char, make_buffer_string, make_buffer_string_both)
(update_buffer_properties, Fbuffer_substring)
(Fbuffer_substring_no_properties, Fcompare_buffer_substrings)
(Fsubst_char_in_region, check_translation)
(Ftranslate_region_internal, save_restriction_restore, Fformat)
(transpose_markers, Ftranspose_regions):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(clip_to_bounds): Move to lisp.h as an inline function).
(Fconstrain_to_field): Don't assume integers are nonnegative.
(Fline_beginning_position, Fsave_excursion, Fsave_current_buffer):
(Fsubst_char_in_region, Fsave_restriction):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Femacs_pid): Don't assume pid_t fits into fixnum.
(lo_time): Use int, not EMACS_INT, when int suffices.
(lisp_time_argument): Check for usec out of range.
(Fencode_time): Don't assume fixnum fits in int.
* emacs.c (gdb_valbits, gdb_gctypebits): Now int, not EMACS_INT.
(gdb_data_seg_bits): Now uintptr_t, not EMACS_INT.
(PVEC_FLAG, gdb_array_mark_flag): Now ptrdiff_t, not EMACS_INT.
(init_cmdargs, Fdump_emacs):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Fkill_emacs): Don't assume fixnum fits in int; instead, take just
the bottom (typically) 32 bits of the fixnum.
* eval.c (specpdl_size, call_debugger):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(when_entered_debugger, Fbacktrace_debug):
Don't assume fixnum can fit in int.
(Fdefvaralias, Fdefvar): Do not attempt to compute the address of
the object just before a buffer; this is not portable.
(FletX, Flet, Funwind_protect, do_autoload, Feval, funcall_lambda)
(grow_specpdl, unbind_to):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Fapply, apply_lambda): Don't assume ptrdiff_t can hold fixnum.
(grow_specpdl): Simplify allocation by using xpalloc.
* fileio.c (Ffind_file_name_handler, Fcopy_file, Frename_file)
(Finsert_file_contents, Fwrite_region, Fdo_auto_save):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Ffind_file_name_handler, non_regular_inserted, Finsert_file_contents)
(a_write, e_write):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fcopy_file, non_regular_nbytes, read_non_regular)
(Finsert_file_contents):
Use int, not EMACS_INT, where int is wide enough.
(READ_BUF_SIZE): Verify that it fits in int.
(Finsert_file_contents): Check that counts are in proper range,
rather than assuming fixnums fit into ptrdiff_t etc.
Don't assume fixnums fit into int.
(Fdo_auto_save, Fset_buffer_auto_saved)
(Fclear_buffer_auto_save_failure):
Don't assume time_t is signed, or that it fits in int.
* fns.c (Fcompare_strings, Fstring_lessp, struct textprop_rec)
(concat, string_char_byte_cache_charpos, string_char_byte_cache_bytepos)
(string_char_to_byte, string_byte_to_char)
(string_make_multibyte, string_to_multibyte)
(string_make_unibyte, Fstring_as_unibyte, Fstring_as_multibyte)
(Fstring_to_unibyte, Fsubstring, Fsubstring_no_properties)
(substring_both, Fdelete, internal_equal, Ffillarray)
(Fclear_string, mapcar1)
(Fbase64_encode_region, Fbase64_encode_string, base64_encode_1)
(Fbase64_decode_region, Fbase64_decode_string, base64_decode_1)
(larger_vector, make_hash_table, maybe_resize_hash_table)
(hash_lookup, hash_remove_from_table, hash_clear, sweep_weak_table)
(Fmaphash, secure_hash):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(concat): Check for string index and length overflow.
(Fmapconcat): Don't assume fixnums fit into ptrdiff_t.
(Frequire):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(larger_vector): New API (vec, incr_min, size_max) replaces old
one (vec, new_size, init).  This catches size overflow.
INIT was removed because it was always Qnil.
All callers changed.
(INDEX_SIZE_BOUND): New macro, which calculates more precisely
the upper bound on a hash table index size.
(make_hash_table, maybe_resize_hash_table): Use it.
(secure_hash): Computer start_byte and end_byte only after
they're known to be in ptrdiff_t range.
* font.c (font_intern_prop, font_at, font_range, Ffont_shape_gstring)
(Ffont_get_glyphs, Ffont_at):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(font_style_to_value, font_prop_validate_style, font_expand_wildcards)
(Flist_fonts, Fopen_font):
Don't assume fixnum can fit in int.
(check_gstring): Don't assume index can fit in int.
(font_match_p): Check that fixnum is a character, not a nonnegative
fixnum, since the later code needs to stuff it into an int.
(font_find_for_lface): Use SAFE_ALLOCA_LISP, not alloca.
(font_fill_lglyph_metrics): Use unsigned, not EMACS_INT, to avoid
conversion overflow issues.
(Fopen_font): Check for integer out of  range.
(Ffont_get_glyphs): Don't assume index can fit in int.
* font.h: Adjust decls to match defn changes elsewhere.
* fontset.c (reorder_font_vector): Redo score calculation to avoid
integer overflow.
(num_auto_fontsets, fontset_from_font): Use ptrdiff_t, not
printmax_t, where ptrdiff_t is wide enough.
(Finternal_char_font):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* frame.c (Fset_mouse_position, Fset_mouse_pixel_position)
(Fset_frame_height, Fset_frame_width, Fset_frame_size)
(Fset_frame_position, x_set_frame_parameters)
(x_set_line_spacing, x_set_border_width)
(x_set_internal_border_width, x_set_alpha, x_figure_window_size):
Check that fixnums are in proper range for system types.
(frame_name_fnn_p, Fframe_parameter, Fmodify_frame_parameters):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fmodify_frame_parameters): Don't assume fixnum fits in int.
Use SAFE_ALLOCA_LISP, not alloca.
* frame.h (struct frame): Use intptr_t, not EMACS_INT, where
intptr_t is wide enough.
* fringe.c (lookup_fringe_bitmap, get_logical_fringe_bitmap)
(Fdefine_fringe_bitmap): Don't assume fixnum fits in int.
(Ffringe_bitmaps_at_pos): Don't assume index fits in int.
Check for fixnum out of range.
* ftfont.c (ftfont_list): Don't assume index fits in int.
Check that fixnums are in proper range for system types.
(ftfont_shape_by_flt):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
Remove no-longer-needed lint_assume.
* gnutls.c (emacs_gnutls_write, emacs_gnutls_read):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fgnutls_error_fatalp, Fgnutls_error_string, Fgnutls_boot):
Check that fixnums are in proper range for system types.
* gnutls.h: Adjust decls to match defn changes elsewhere.
* gtkutil.c (xg_dialog_run):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(update_frame_tool_bar):
Check that fixnums are in proper range for system types.
* image.c (parse_image_spec): Redo count calculation to avoid overflow.
(lookup_image): Check that fixnums are in proper range for system types.
* indent.c (last_known_column, last_known_column_point):
(current_column_bol_cache):
(skip_invisible, current_column, check_display_width):
(check_display_width, scan_for_column, current_column_1)
(Findent_to, Fcurrent_indentation, position_indentation)
(indented_beyond_p, Fmove_to_column, compute_motion):
(Fcompute_motion, Fvertical_motion):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(last_known_column_modified): Use EMACS_INT, not int.
(check_display_width):
(Fcompute_motion):
Check that fixnums and floats are in proper range for system types.
(compute_motion): Don't assume index or fixnum fits in int.
(compute_motion, Fcompute_motion):
Use int, not EMACS_INT, when it is wide enough.
(vmotion): Omit local var start_hpos that is always 0; that way
we don't need to worry about overflow in expressions involving it.
* indent.h: Adjust decls to match defn changes elsewhere.
(struct position):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
Use int, not EMACS_INT, where int is wide enough.
Remove unused members ovstring_chars_done and tab_offset;
all uses removed.
* insdel.c (move_gap, move_gap_both, gap_left, gap_right)
(adjust_markers_for_delete, adjust_markers_for_insert, adjust_point)
(adjust_markers_for_replace, make_gap_larger, make_gap_smaller)
(make_gap, copy_text, insert, insert_and_inherit)
(insert_before_markers, insert_before_markers_and_inherit)
(insert_1, count_combining_before, count_combining_after)
(insert_1_both, insert_from_string)
(insert_from_string_before_markers, insert_from_string_1)
(insert_from_gap, insert_from_buffer, insert_from_buffer_1)
(adjust_after_replace, adjust_after_insert, replace_range)
(replace_range_2, del_range, del_range_1, del_range_byte)
(del_range_both, del_range_2, modify_region)
(prepare_to_modify_buffer, signal_before_change)
(signal_after_change, Fcombine_after_change_execute):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* intervals.c (traverse_intervals, rotate_right, rotate_left)
(balance_an_interval, split_interval_right, split_interval_left)
(find_interval, next_interval, update_interval)
(adjust_intervals_for_insertion, delete_node, delete_interval)
(interval_deletion_adjustment, adjust_intervals_for_deletion)
(static_offset_intervals, offset_intervals)
(merge_interval_right, merge_interval_left, make_new_interval)
(graft_intervals_into_buffer, temp_set_point_both)
(temp_set_point, set_point, adjust_for_invis_intang)
(set_point_both, move_if_not_intangible, get_property_and_range)
(get_local_map, copy_intervals, copy_intervals_to_string)
(compare_string_intervals, set_intervals_multibyte_1):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* intervals.h: Adjust decls to match defn changes elsewhere.
(struct interval):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* keyboard.c (this_command_key_count, this_single_command_key_start)
(before_command_key_count, before_command_echo_length, echo_now)
(echo_length, recursive_edit_1, Frecursive_edit, Ftrack_mouse)
(command_loop_1, safe_run_hooks, read_char, timer_check_2)
(menu_item_eval_property, read_key_sequence, Fread_key_sequence)
(Fread_key_sequence_vector, Fexecute_extended_command, Fsuspend_emacs):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(last_non_minibuf_size, last_point_position, echo_truncate)
(command_loop_1, adjust_point_for_property, read_char, gen_help_event)
(make_lispy_position, make_lispy_event, parse_modifiers_uncached)
(parse_modifiers, modify_event_symbol, Fexecute_extended_command)
(stuff_buffered_input):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(last_auto_save, command_loop_1, read_char):
Use EMACS_INT, not int, to avoid integer overflow.
(record_char): Avoid overflow in total_keys computation.
(parse_modifiers_uncached): Redo index calculation to avoid overflow.
* keyboard.h: Adjust decls to match defn changes elsewhere.
* keymap.c (Fdefine_key, Fcurrent_active_maps, accessible_keymaps_1)
(Fkey_description, Fdescribe_vector, Flookup_key):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(click_position): New function, to check that positions are in range.
(Fcurrent_active_maps):
(describe_command):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Faccessible_keymaps, Fkey_description):
(preferred_sequence_p):
Don't assume fixnum can fit into int.
(Fkey_description): Use SAFE_ALLOCA_LISP, not alloca.
Check for integer overflow in size calculations.
(Ftext_char_description): Use CHECK_CHARACTER, not CHECK_NUMBER, to
avoid mishandling large integers.
* lisp.h: Adjust decls to match defn changes elsewhere.
(ARRAY_MARK_FLAG, PSEUDOVECTOR_FLAG, struct Lisp_String)
(struct vectorlike_header, struct Lisp_Subr, struct Lisp_Hash_Table)
(struct Lisp_Marker):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(clip_to_bounds): Now an inline function, moved here from editfns.c.
(XSETSUBR): Use size of 0 since the actual size doesn't matter,
and using 0 avoids overflow.
(GLYPH_CODE_P): Check for overflow in system types, subsuming the
need for GLYPH_CODE_CHAR_VALID_P and doing proper checking ourselves.
All callers changed.
(GLYPH_CODE_CHAR, GLYPH_CODE_FACE):
Assume the arg has valid form, since it always does.
(TYPE_RANGED_INTEGERP): Avoid bug when checking against a wide
unsigned integer system type.
(CHECK_RANGED_INTEGER, CHECK_TYPE_RANGED_INTEGER): New macros.
(struct catchtag, specpdl_size, SPECPDL_INDEX, USE_SAFE_ALLOCA):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(struct catchtag): Use EMACS_INT, not int, since it may be a fixnum.
(duration_to_sec_usec): New decl.
* lread.c (read_from_string_index, read_from_string_index_byte)
(read_from_string_limit, readchar, unreadchar, openp)
(read_internal_start, read1, oblookup):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fload, readevalloop, Feval_buffer, Feval_region):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(openp): Check for out-of-range argument to 'access'.
(read1): Use int, not EMACS_INT, where int is wide enough.
Don't assume fixnum fits into int.
* macros.c (Fstart_kbd_macro): Use xpalloc to check for overflow
in size calculation.
(Fexecute_kbd_macro):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* marker.c (cached_charpos, cached_bytepos, CONSIDER)
(byte_char_debug_check, buf_charpos_to_bytepos, verify_bytepos)
(buf_bytepos_to_charpos, Fset_marker, set_marker_restricted)
(set_marker_both, set_marker_restricted_both, marker_position)
(marker_byte_position, Fbuffer_has_markers_at):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fset_marker, set_marker_restricted): Don't assume fixnum fits in int.
* menu.c (ensure_menu_items): Renamed from grow_menu_items.
It now merely ensures that the menu is large enough, without
necessarily growing it, as this avoids some integer overflow issues.
All callers changed.
(keymap_panes, parse_single_submenu, Fx_popup_menu):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(parse_single_submenu, Fx_popup_menu): Don't assume fixnum fits in int.
Use SAFE_ALLOCA_LISP, not alloca.
(find_and_return_menu_selection): Avoid unnecessary casts of pointers
to EMACS_INT.  Check that fixnums are in proper range for system types.
* minibuf.c (minibuf_prompt_width, string_to_object)
(Fminibuffer_contents, Fminibuffer_contents_no_properties)
(Fminibuffer_completion_contents, Ftry_completion, Fall_completions):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(get_minibuffer, read_minibuf_unwind):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(read_minibuf): Omit unnecessary arg BACKUP_N, which is always nil;
this simplifies overflow checking.  All callers changed.
(read_minibuf, Fread_buffer, Ftry_completion, Fall_completions)
(Ftest_completion):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* nsfns.m (check_ns_display_info): Don't assume fixnum fits in long.
(x_set_menu_bar_lines, x_set_tool_bar_lines, Fx_create_frame):
Check that fixnums are in proper range for system types.
(Fx_create_frame, Fx_show_tip):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* nsfont.m (ns_findfonts, nsfont_list_family):
Don't assume fixnum fits in long.
* nsmenu.m (ns_update_menubar, ns_menu_show, ns_popup_dialog):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(ns_update_menubar): Use intptr_t, not EMACS_INT, when intptr_t is
wide enough.
* nsselect.m (ns_get_local_selection):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* print.c (print_buffer_size, print_buffer_pos, print_buffer_pos_byte)
(PRINTDECLARE, PRINTPREPARE):
(strout, print_string):
(print, print_preprocess, print_check_string_charset_prop)
(print_object):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(PRINTDECLARE):
(temp_output_buffer_setup, Fprin1_to_string, print_object):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(PRINTPREPARE): Use int, not ptrdiff_t, where int is wide enough.
(PRINTFINISH): Use SAFE_ALLOCA, not alloca.
(printchar, strout): Use xpalloc to catch size calculation overflow.
(Fexternal_debugging_output): Use CHECK_CHARACTER, not CHECK_NUMBER,
to avoid mishandling large integers.
(print_error_message): Use SAFE_ALLOCA, not alloca.
(print_object): Use int, not EMACS_INT, where int is wide enough.
* process.c (Fdelete_process): Don't assume pid fits into EMACS_INT.
(Fset_process_window_size, Fformat_network_address)
(get_lisp_to_sockaddr_size, set_socket_option, Fmake_network_process)
(Fsignal_process, sigchld_handler):
Check that fixnums are in proper range for system types.
(Fformat_network_address, read_process_output, send_process)
(Fprocess_send_region, status_notify):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fformat_network_address, Fmake_serial_process, Fmake_network_process)
(wait_reading_process_output, read_process_output, exec_sentinel):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(conv_lisp_to_sockaddr): Don't assume fixnums fit into int.
(Faccept_process_output): Use duration_to_sec_usec to do proper
overflow checking on durations.
* scroll.c (calculate_scrolling, calculate_direct_scrolling)
(line_ins_del): Use int, not EMACS_INT, where int is wide enough.
* search.c (looking_at_1, string_match_1):
(fast_string_match, fast_c_string_match_ignore_case)
(fast_string_match_ignore_case, fast_looking_at, scan_buffer)
(scan_newline, find_before_next_newline, search_command)
(trivial_regexp_p, search_buffer, simple_search, boyer_moore)
(set_search_regs, wordify):
(Freplace_match):
(Fmatch_data):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(string_match_1, search_buffer, set_search_regs):
(Fmatch_data):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(wordify): Check for overflow in size calculation.
(Freplace_match): Avoid potential buffer overflow in search_regs.start.
(Fset_match_data): Don't assume fixnum fits in ptrdiff_t.
Check that fixnums are in proper range for system types.
* sound.c (struct sound_device)
(wav_play, au_play, vox_write, alsa_period_size, alsa_write):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fplay_sound_internal):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* syntax.c (ST_COMMENT_STYLE, ST_STRING_STYLE):
In definitions, make it clearer that these values must be out of range
for the respective integer ranges.  This fixes a bug with ST_STRING_STYLE
and non-ASCII characters.
(struct lisp_parse_state, find_start_modiff)
(Finternal_describe_syntax_value, scan_lists, scan_sexps_forward):
(Fparse_partial_sexp):
Don't assume fixnums can fit in int.
(struct lisp_parse_state, find_start_pos, find_start_value)
(find_start_value_byte, find_start_begv)
(update_syntax_table, char_quoted, dec_bytepos)
(find_defun_start, prev_char_comend_first, back_comment):
(scan_words, skip_chars, skip_syntaxes, forw_comment, Fforward_comment)
(scan_lists, Fbackward_prefix_chars, scan_sexps_forward):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Finternal_describe_syntax_value): Check that match_lisp is a
character, not an integer, since the code stuffs it into int.
(scan_words, scan_sexps_forward):
Check that fixnums are in proper range for system types.
(Fforward_word):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(scan_sexps_forward):
Use CHARACTERP, not INTEGERP, since the value must fit into int.
(Fparse_partial_sexp): Fix doc; element 8 is not ignored.
* syntax.h: Adjust decls to match defn changes elsewhere.
(struct gl_state_s):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* sysdep.c (wait_for_termination_1, wait_for_termination)
(interruptible_wait_for_termination, mkdir):
Don't assume pid_t fits in int; on 64-bit AIX pid_t is 64-bit.
(emacs_read, emacs_write):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(system_process_attributes): Don't assume uid_t, gid_t, and
double all fit in int or even EMACS_INT.
* term.c (set_tty_color_mode):
Check that fixnums are in proper range for system types.
* termhooks.h (struct input_event):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* textprop.c (validate_interval_range, interval_of)
(Fadd_text_properties, set_text_properties_1)
(Fremove_text_properties, Fremove_list_of_text_properties)
(Ftext_property_any, Ftext_property_not_all)
(copy_text_properties, text_property_list, extend_property_ranges)
(verify_interval_modification):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fnext_single_char_property_change)
(Fprevious_single_char_property_change):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(copy_text_properties): Check for integer overflow in index calculation.
* undo.c (last_boundary_position, record_point, record_insert)
(record_delete, record_marker_adjustment, record_change)
(record_property_change):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(truncate_undo_list, Fprimitive_undo): Don't assume fixnum fits in int.
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* w32fns.c (Fx_create_frame, x_create_tip_frame, Fx_show_tip)
(Fx_hide_tip, Fx_file_dialog):
* w32menu.c (set_frame_menubar):
Use ptrdiff_t, not int, for consistency with rest of code.
* window.c (window_scroll_preserve_hpos, window_scroll_preserve_vpos)
(select_window, Fdelete_other_windows_internal)
(window_scroll_pixel_based, window_scroll_line_based)
(Frecenter, Fset_window_configuration):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(Fset_window_hscroll, run_window_configuration_change_hook)
(set_window_buffer, temp_output_buffer_show, scroll_command)
(Fscroll_other_window):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Fwindow_line_height, window_scroll, Fscroll_left, Fscroll_right):
Don't assume fixnum fits in int.
(Fset_window_scroll_bars):
Check that fixnums are in proper range for system types.
* xdisp.c (help_echo_pos, pos_visible_p, string_pos_nchars_ahead)
(string_pos, c_string_pos, number_of_chars, init_iterator)
(in_ellipses_for_invisible_text_p, init_from_display_pos)
(compute_stop_pos, next_overlay_change, compute_display_string_pos)
(compute_display_string_end, handle_face_prop)
(face_before_or_after_it_pos, handle_invisible_prop, handle_display_prop)
(handle_display_spec, handle_single_display_spec)
(display_prop_intangible_p, string_buffer_position_lim)
(string_buffer_position, handle_composition_prop, load_overlay_strings)
(get_overlay_strings_1, get_overlay_strings)
(iterate_out_of_display_property, forward_to_next_line_start)
(back_to_previous_visible_line_start, reseat, reseat_to_string)
(get_next_display_element, set_iterator_to_next)
(get_visually_first_element, compute_stop_pos_backwards)
(handle_stop_backwards, next_element_from_buffer)
(move_it_in_display_line_to, move_it_in_display_line)
(move_it_to, move_it_vertically_backward, move_it_by_lines)
(add_to_log, message_dolog, message_log_check_duplicate)
(message2, message2_nolog, message3, message3_nolog
(with_echo_area_buffer, display_echo_area_1, resize_mini_window_1)
(current_message_1, truncate_echo_area, truncate_message_1)
(set_message, set_message_1, store_mode_line_noprop)
(hscroll_window_tree, debug_delta, debug_delta_bytes, debug_end_vpos)
(text_outside_line_unchanged_p, check_point_in_composition)
(reconsider_clip_changes)
(redisplay_internal, set_cursor_from_row, try_scrolling)
(try_cursor_movement, set_vertical_scroll_bar, redisplay_window)
(redisplay_window, find_last_unchanged_at_beg_row)
(find_first_unchanged_at_end_row, row_containing_pos, try_window_id)
(trailing_whitespace_p, find_row_edges, display_line)
(RECORD_MAX_MIN_POS, Fcurrent_bidi_paragraph_direction)
(display_mode_element, store_mode_line_string)
(pint2str, pint2hrstr, decode_mode_spec)
(display_count_lines, display_string, draw_glyphs)
(x_produce_glyphs, x_insert_glyphs)
(rows_from_pos_range, mouse_face_from_buffer_pos)
(fast_find_string_pos, mouse_face_from_string_pos)
(note_mode_line_or_margin_highlight, note_mouse_highlight):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(safe_call, init_from_display_pos, handle_fontified_prop)
(handle_single_display_spec, load_overlay_strings)
(with_echo_area_buffer, setup_echo_area_for_printing)
(display_echo_area, echo_area_display)
(x_consider_frame_title, prepare_menu_bars, update_menu_bar)
(update_tool_bar, hscroll_window_tree, redisplay_internal)
(redisplay_window, dump_glyph_row, display_mode_line, Fformat_mode_line)
(decode_mode_spec, on_hot_spot_p):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(handle_single_display_spec, build_desired_tool_bar_string)
(redisplay_tool_bar, scroll_window_tree, Fdump_glyph_matrix)
(get_specified_cursor_type):
Check that fixnums are in proper range for system types.
(struct overlay_entry, resize_mini_window, Fdump_glyph_row)
(Flookup_image_map):
Don't assume fixnums fit in int.
(compare_overlay_entries):
Avoid mishandling comparisons due to subtraction overflow.
(load_overlay_strings): Use SAFE_NALLOCA, not alloca.
(last_escape_glyph_face_id, last_glyphless_glyph_face_id):
(handle_tool_bar_click):
Use int, not unsigned, since we prefer signed and the signedness
doesn't matter here.
(get_next_display_element, next_element_from_display_vector):
Use int, not EMACS_INT, when int is wide enough.
(start_hourglass): Use duration_to_sec_usec to do proper
overflow checking on durations.
* xfaces.c (Fbitmap_spec_p):
Check that fixnums are in proper range for system types.
(compare_fonts_by_sort_order):
Avoid mishandling comparisons due to subtraction overflow.
(Fx_family_fonts, realize_basic_faces):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Fx_family_fonts):
Don't assume fixnum fits in int.
Use SAFE_ALLOCA_LISP, not alloca.
(merge_face_heights): Remove unnecessary cast to EMACS_INT.
(Finternal_make_lisp_face): Don't allocate more than MAX_FACE_ID.
(face_at_buffer_position, face_for_overlay_string)
(face_at_string_position):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
(merge_faces): Use int, not EMACS_INT, where int is wide enough.
* xfns.c (x_set_menu_bar_lines, x_set_tool_bar_lines, x_icon_verify)
(Fx_show_tip):
Check that fixnums are in proper range for system types.
(Fx_create_frame, x_create_tip_frame, Fx_show_tip)
(Fx_hide_tip, Fx_file_dialog, Fx_select_font):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(Fx_change_window_property): Don't assume fixnums fit in int.
* xfont.c (xfont_chars_supported):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* xmenu.c (Fx_popup_dialog, set_frame_menubar)
(create_and_show_popup_menu, create_and_show_dialog, xmenu_show):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* xml.c (parse_region):
* xrdb.c (magic_file_p):
Use ptrdiff_t, not EMACS_INT, where ptrdiff_t is wide enough.
* xselect.c (TRACE1): Don't assume pid_t promotes to int.
(x_get_local_selection, x_reply_selection_request)
(x_handle_selection_request, wait_for_property_change):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
(selection_data_to_lisp_data): Use short, not EMACS_INT, where
short is wide enough.
(x_send_client_event): Don't assume fixnum fits in int.
* xterm.c (x_x_to_emacs_modifiers):
Don't assume EMACS_INT overflows nicely into int.
(x_emacs_to_x_modifiers): Use EMACS_INT, not int, because values
may come from Lisp.
(handle_one_xevent): NATNUMP can eval its arg twice.
(x_connection_closed):
Use ptrdiff_t, not int, to avoid needless 32-bit limit on 64-bit hosts.
* xterm.h: Adjust decls to match defn changes elsewhere.
(struct scroll_bar): Use struct vectorlike_header
rather than rolling our own approximation.
(SCROLL_BAR_VEC_SIZE): Remove; not used.
2011-09-21 10:41:20 -07:00

991 lines
32 KiB
C
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

/* Calculate what line insertion or deletion to do, and do it
Copyright (C) 1985-1986, 1990, 1993-1994, 2001-2011
Free Software Foundation, Inc.
This file is part of GNU Emacs.
GNU Emacs is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
GNU Emacs is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */
#include <config.h>
#include <stdio.h>
#include <setjmp.h>
#include "lisp.h"
#include "termchar.h"
#include "dispextern.h"
#include "keyboard.h"
#include "frame.h"
#include "window.h"
#include "termhooks.h"
/* All costs measured in characters.
So no cost can exceed the area of a frame, measured in characters.
Let's hope this is never more than 1000000 characters. */
#define INFINITY 1000000
struct matrix_elt
{
/* Cost of outputting through this line
if no insert/delete is done just above it. */
int writecost;
/* Cost of outputting through this line
if an insert is done just above it. */
int insertcost;
/* Cost of outputting through this line
if a delete is done just above it. */
int deletecost;
/* Number of inserts so far in this run of inserts,
for the cost in insertcost. */
unsigned char insertcount;
/* Number of deletes so far in this run of deletes,
for the cost in deletecost. */
unsigned char deletecount;
/* Number of writes so far since the last insert
or delete for the cost in writecost. */
unsigned char writecount;
};
static void do_direct_scrolling (struct frame *,
struct glyph_matrix *,
struct matrix_elt *,
int, int);
static void do_scrolling (struct frame *,
struct glyph_matrix *,
struct matrix_elt *,
int, int);
/* Determine, in matrix[i,j], the cost of updating the first j old
lines into the first i new lines using the general scrolling method.
This involves using insert or delete somewhere if i != j.
For each matrix elements, three kinds of costs are recorded:
the smallest cost that ends with an insert, the smallest
cost that ends with a delete, and the smallest cost that
ends with neither one. These are kept separate because
on some terminals the cost of doing an insert varies
depending on whether one was just done, etc. */
/* draw_cost[VPOS] is the cost of outputting new line at VPOS.
old_hash[VPOS] is the hash code of the old line at VPOS.
new_hash[VPOS] is the hash code of the new line at VPOS.
Note that these are not true frame vpos's, but relative
to the place at which the first mismatch between old and
new contents appears. */
static void
calculate_scrolling (FRAME_PTR frame,
/* matrix is of size window_size + 1 on each side. */
struct matrix_elt *matrix,
int window_size, int lines_below,
int *draw_cost, int *old_hash, int *new_hash,
int free_at_end)
{
register int i, j;
int frame_lines = FRAME_LINES (frame);
register struct matrix_elt *p, *p1;
register int cost, cost1;
int lines_moved = window_size
+ (FRAME_SCROLL_REGION_OK (frame) ? 0 : lines_below);
/* first_insert_cost[I] is the cost of doing the first insert-line
at the i'th line of the lines we are considering,
where I is origin 1 (as it is below). */
int *first_insert_cost
= &FRAME_INSERT_COST (frame)[frame_lines - 1 - lines_moved];
int *first_delete_cost
= &FRAME_DELETE_COST (frame)[frame_lines - 1 - lines_moved];
int *next_insert_cost
= &FRAME_INSERTN_COST (frame)[frame_lines - 1 - lines_moved];
int *next_delete_cost
= &FRAME_DELETEN_COST (frame)[frame_lines - 1 - lines_moved];
/* Discourage long scrolls on fast lines.
Don't scroll nearly a full frame height unless it saves
at least 1/4 second. */
int extra_cost = (int) (baud_rate / (10 * 4 * FRAME_LINES (frame)));
if (baud_rate <= 0)
extra_cost = 1;
/* initialize the top left corner of the matrix */
matrix->writecost = 0;
matrix->insertcost = INFINITY;
matrix->deletecost = INFINITY;
matrix->insertcount = 0;
matrix->deletecount = 0;
/* initialize the left edge of the matrix */
cost = first_insert_cost[1] - next_insert_cost[1];
for (i = 1; i <= window_size; i++)
{
p = matrix + i * (window_size + 1);
cost += draw_cost[i] + next_insert_cost[i] + extra_cost;
p->insertcost = cost;
p->writecost = INFINITY;
p->deletecost = INFINITY;
p->insertcount = i;
p->deletecount = 0;
}
/* initialize the top edge of the matrix */
cost = first_delete_cost[1] - next_delete_cost[1];
for (j = 1; j <= window_size; j++)
{
cost += next_delete_cost[j];
matrix[j].deletecost = cost;
matrix[j].writecost = INFINITY;
matrix[j].insertcost = INFINITY;
matrix[j].deletecount = j;
matrix[j].insertcount = 0;
}
/* `i' represents the vpos among new frame contents.
`j' represents the vpos among the old frame contents. */
p = matrix + window_size + 2; /* matrix [1, 1] */
for (i = 1; i <= window_size; i++, p++)
for (j = 1; j <= window_size; j++, p++)
{
/* p contains the address of matrix [i, j] */
/* First calculate the cost assuming we do
not insert or delete above this line.
That is, if we update through line i-1
based on old lines through j-1,
and then just change old line j to new line i. */
p1 = p - window_size - 2; /* matrix [i-1, j-1] */
cost = p1->writecost;
if (cost > p1->insertcost)
cost = p1->insertcost;
if (cost > p1->deletecost)
cost = p1->deletecost;
if (old_hash[j] != new_hash[i])
cost += draw_cost[i];
p->writecost = cost;
/* Calculate the cost if we do an insert-line
before outputting this line.
That is, we update through line i-1
based on old lines through j,
do an insert-line on line i,
and then output line i from scratch,
leaving old lines starting from j for reuse below. */
p1 = p - window_size - 1; /* matrix [i-1, j] */
/* No need to think about doing a delete followed
immediately by an insert. It cannot be as good
as not doing either of them. */
if (free_at_end == i)
{
cost = p1->writecost;
cost1 = p1->insertcost;
}
else
{
cost = p1->writecost + first_insert_cost[i];
if ((int) p1->insertcount > i)
abort ();
cost1 = p1->insertcost + next_insert_cost[i - p1->insertcount];
}
p->insertcost = min (cost, cost1) + draw_cost[i] + extra_cost;
p->insertcount = (cost < cost1) ? 1 : p1->insertcount + 1;
if ((int) p->insertcount > i)
abort ();
/* Calculate the cost if we do a delete line after
outputting this line.
That is, we update through line i
based on old lines through j-1,
and throw away old line j. */
p1 = p - 1; /* matrix [i, j-1] */
/* No need to think about doing an insert followed
immediately by a delete. */
if (free_at_end == i)
{
cost = p1->writecost;
cost1 = p1->deletecost;
}
else
{
cost = p1->writecost + first_delete_cost[i];
cost1 = p1->deletecost + next_delete_cost[i];
}
p->deletecost = min (cost, cost1);
p->deletecount = (cost < cost1) ? 1 : p1->deletecount + 1;
}
}
/* Perform insert-lines and delete-lines operations on CURRENT_MATRIX
according to the costs in MATRIX, using the general scrolling
method that is used if the terminal does not support the setting of
scroll windows (scroll_region_ok == 0).
WINDOW_SIZE is the number of lines being considered for scrolling
and UNCHANGED_AT_TOP is the vpos of the first line being
considered. These two arguments can specify any contiguous range
of lines. */
static void
do_scrolling (struct frame *frame, struct glyph_matrix *current_matrix,
struct matrix_elt *matrix, int window_size,
int unchanged_at_top)
{
struct matrix_elt *p;
int i, j, k;
/* Set to 1 if we have set a terminal window with
set_terminal_window. It's unsigned to work around GCC bug 48228. */
unsigned int terminal_window_p = 0;
/* A queue for line insertions to be done. */
struct queue { int count, pos; };
struct queue *queue_start
= (struct queue *) alloca (current_matrix->nrows * sizeof (struct queue));
struct queue *queue = queue_start;
char *retained_p = (char *) alloca (window_size * sizeof (char));
int *copy_from = (int *) alloca (window_size * sizeof (int));
/* Zero means line is empty. */
memset (retained_p, 0, window_size * sizeof (char));
for (k = 0; k < window_size; ++k)
copy_from[k] = -1;
#if GLYPH_DEBUG
# define CHECK_BOUNDS \
do \
{ \
int ck; \
for (ck = 0; ck < window_size; ++ck) \
xassert (copy_from[ck] == -1 \
|| (copy_from[ck] >= 0 && copy_from[ck] < window_size)); \
} \
while (0);
#endif
/* When j is advanced, this corresponds to deleted lines.
When i is advanced, this corresponds to inserted lines. */
i = j = window_size;
while (i > 0 || j > 0)
{
p = matrix + i * (window_size + 1) + j;
if (p->insertcost < p->writecost && p->insertcost < p->deletecost)
{
/* Insert should be done at vpos i-1, plus maybe some before.
Queue the screen operation to be performed. */
queue->count = p->insertcount;
queue->pos = i + unchanged_at_top - p->insertcount;
++queue;
/* By incrementing I, we leave room in the result rows
for the empty rows opened up. */
i -= p->insertcount;
}
else if (p->deletecost < p->writecost)
{
/* Old line at vpos j-1, and maybe some before it, should be
deleted. By decrementing J, we skip some lines in the
temp_rows which is equivalent to omitting these lines in
the result rows, thus deleting them. */
j -= p->deletecount;
/* Set the terminal window, if not done already. */
if (! terminal_window_p)
{
set_terminal_window (frame, window_size + unchanged_at_top);
terminal_window_p = 1;
}
/* Delete lines on the terminal. */
ins_del_lines (frame, j + unchanged_at_top, - p->deletecount);
}
else
{
/* Best thing done here is no insert or delete, i.e. a write. */
--i, --j;
xassert (i >= 0 && i < window_size);
xassert (j >= 0 && j < window_size);
copy_from[i] = j;
retained_p[j] = 1;
#if GLYPH_DEBUG
CHECK_BOUNDS;
#endif
}
}
/* Now do all insertions queued above. */
if (queue > queue_start)
{
int next = -1;
/* Set the terminal window if not yet done. */
if (!terminal_window_p)
{
set_terminal_window (frame, window_size + unchanged_at_top);
terminal_window_p = 1;
}
do
{
--queue;
/* Do the deletion on the terminal. */
ins_del_lines (frame, queue->pos, queue->count);
/* All lines in the range deleted become empty in the glyph
matrix. Assign to them glyph rows that are not retained.
K is the starting position of the deleted range relative
to the window we are working in. */
k = queue->pos - unchanged_at_top;
for (j = 0; j < queue->count; ++j)
{
/* Find the next row not retained. */
while (retained_p[++next])
;
/* Record that this row is to be used for the empty
glyph row j. */
copy_from[k + j] = next;
}
}
while (queue > queue_start);
}
for (k = 0; k < window_size; ++k)
xassert (copy_from[k] >= 0 && copy_from[k] < window_size);
/* Perform the row swizzling. */
mirrored_line_dance (current_matrix, unchanged_at_top, window_size,
copy_from, retained_p);
/* Some sanity checks if GLYPH_DEBUG != 0. */
CHECK_MATRIX (current_matrix);
if (terminal_window_p)
set_terminal_window (frame, 0);
}
/* Determine, in matrix[i,j], the cost of updating the first j
old lines into the first i new lines using the direct
scrolling method. When the old line and the new line have
different hash codes, the calculated cost of updating old
line j into new line i includes the cost of outputting new
line i, and if i != j, the cost of outputting the old line j
is also included, as a penalty for moving the line and then
erasing it. In addition, the cost of updating a sequence of
lines with constant i - j includes the cost of scrolling the
old lines into their new positions, unless i == j. Scrolling
is achieved by setting the screen window to avoid affecting
other lines below, and inserting or deleting lines at the top
of the scrolled region. The cost of scrolling a sequence of
lines includes the fixed cost of specifying a scroll region,
plus a variable cost which can depend upon the number of lines
involved and the distance by which they are scrolled, and an
extra cost to discourage long scrolls.
As reflected in the matrix, an insert or delete does not
correspond directly to the insertion or deletion which is
used in scrolling lines. An insert means that the value of i
has increased without a corresponding increase in the value
of j. A delete means that the value of j has increased
without a corresponding increase in the value of i. A write
means that i and j are both increased by the same amount, and
that the old lines will be moved to their new positions.
An insert following a delete is allowed only if i > j.
A delete following an insert is allowed only if i < j.
These restrictions ensure that the new lines in an insert
will always be blank as an effect of the neighboring writes.
Thus the calculated cost of an insert is simply the cost of
outputting the new line contents. The direct cost of a
delete is zero. Inserts and deletes indirectly affect the
total cost through their influence on subsequent writes. */
/* The vectors draw_cost, old_hash, and new_hash have the same
meanings here as in calculate_scrolling, and old_draw_cost
is the equivalent of draw_cost for the old line contents */
static void
calculate_direct_scrolling (FRAME_PTR frame,
/* matrix is of size window_size + 1 on each side. */
struct matrix_elt *matrix,
int window_size, int lines_below,
int *draw_cost, int *old_draw_cost,
int *old_hash, int *new_hash,
int free_at_end)
{
register int i, j;
int frame_lines = FRAME_LINES (frame);
register struct matrix_elt *p, *p1;
register int cost, cost1, delta;
/* first_insert_cost[-I] is the cost of doing the first insert-line
at a position I lines above the bottom line in the scroll window. */
int *first_insert_cost
= &FRAME_INSERT_COST (frame)[frame_lines - 1];
int *first_delete_cost
= &FRAME_DELETE_COST (frame)[frame_lines - 1];
int *next_insert_cost
= &FRAME_INSERTN_COST (frame)[frame_lines - 1];
int *next_delete_cost
= &FRAME_DELETEN_COST (frame)[frame_lines - 1];
int scroll_overhead;
/* Discourage long scrolls on fast lines.
Don't scroll nearly a full frame height unless it saves
at least 1/4 second. */
int extra_cost = (int) (baud_rate / (10 * 4 * FRAME_LINES (frame)));
if (baud_rate <= 0)
extra_cost = 1;
/* Overhead of setting the scroll window, plus the extra cost
cost of scrolling by a distance of one. The extra cost is
added once for consistency with the cost vectors */
scroll_overhead
= FRAME_SCROLL_REGION_COST (frame) + extra_cost;
/* initialize the top left corner of the matrix */
matrix->writecost = 0;
matrix->insertcost = INFINITY;
matrix->deletecost = INFINITY;
matrix->writecount = 0;
matrix->insertcount = 0;
matrix->deletecount = 0;
/* initialize the left edge of the matrix */
cost = 0;
for (i = 1; i <= window_size; i++)
{
p = matrix + i * (window_size + 1);
cost += draw_cost[i];
p->insertcost = cost;
p->writecost = INFINITY;
p->deletecost = INFINITY;
p->insertcount = i;
p->writecount = 0;
p->deletecount = 0;
}
/* initialize the top edge of the matrix */
for (j = 1; j <= window_size; j++)
{
matrix[j].deletecost = 0;
matrix[j].writecost = INFINITY;
matrix[j].insertcost = INFINITY;
matrix[j].deletecount = j;
matrix[j].writecount = 0;
matrix[j].insertcount = 0;
}
/* `i' represents the vpos among new frame contents.
`j' represents the vpos among the old frame contents. */
p = matrix + window_size + 2; /* matrix [1, 1] */
for (i = 1; i <= window_size; i++, p++)
for (j = 1; j <= window_size; j++, p++)
{
/* p contains the address of matrix [i, j] */
/* First calculate the cost assuming we do
not insert or delete above this line.
That is, if we update through line i-1
based on old lines through j-1,
and then just change old line j to new line i.
Depending on which choice gives the lower cost,
this usually involves either scrolling a single line
or extending a sequence of scrolled lines, but
when i == j, no scrolling is required. */
p1 = p - window_size - 2; /* matrix [i-1, j-1] */
cost = p1->insertcost;
if (cost > p1->deletecost)
cost = p1->deletecost;
cost1 = p1->writecost;
if (i == j)
{
if (cost > cost1)
{
cost = cost1;
p->writecount = p1->writecount + 1;
}
else
p->writecount = 1;
if (old_hash[j] != new_hash[i])
{
cost += draw_cost[i];
}
}
else
{
if (i > j)
{
delta = i - j;
/* The cost added here for scrolling the first line by
a distance N includes the overhead of setting the
scroll window, the cost of inserting N lines at a
position N lines above the bottom line of the window,
and an extra cost which is proportional to N. */
cost += scroll_overhead + first_insert_cost[-delta] +
(delta-1) * (next_insert_cost[-delta] + extra_cost);
/* In the most general case, the insertion overhead and
the multiply factor can grow linearly as the distance
from the bottom of the window increases. The incremental
cost of scrolling an additional line depends upon the
rate of change of these two parameters. Each of these
growth rates can be determined by a simple difference.
To reduce the cumulative effects of rounding error, we
vary the position at which the difference is computed. */
cost1 += first_insert_cost[-j] - first_insert_cost[1-j] +
(delta-1) * (next_insert_cost[-j] - next_insert_cost[1-j]);
}
else
{
delta = j - i;
cost += scroll_overhead + first_delete_cost[-delta] +
(delta-1) * (next_delete_cost[-delta] + extra_cost);
cost1 += first_delete_cost[-i] - first_delete_cost[1-i] +
(delta-1) * ( next_delete_cost[-i] - next_delete_cost[1-i]);
}
if (cost1 < cost)
{
cost = cost1;
p->writecount = p1->writecount + 1;
}
else
p->writecount = 1;
if (old_hash[j] != new_hash[i])
{
cost += draw_cost[i] + old_draw_cost[j];
}
}
p->writecost = cost;
/* Calculate the cost if we do an insert-line
before outputting this line.
That is, we update through line i-1
based on old lines through j,
do an insert-line on line i,
and then output line i from scratch,
leaving old lines starting from j for reuse below. */
p1 = p - window_size - 1; /* matrix [i-1, j] */
cost = p1->writecost;
/* If i > j, an insert is allowed after a delete. */
if (i > j && p1->deletecost < cost)
cost = p1->deletecost;
if (p1->insertcost <= cost)
{
cost = p1->insertcost;
p->insertcount = p1->insertcount + 1;
}
else
p->insertcount = 1;
cost += draw_cost[i];
p->insertcost = cost;
/* Calculate the cost if we do a delete line after
outputting this line.
That is, we update through line i
based on old lines through j-1,
and throw away old line j. */
p1 = p - 1; /* matrix [i, j-1] */
cost = p1->writecost;
/* If i < j, a delete is allowed after an insert. */
if (i < j && p1->insertcost < cost)
cost = p1->insertcost;
cost1 = p1->deletecost;
if (p1->deletecost <= cost)
{
cost = p1->deletecost;
p->deletecount = p1->deletecount + 1;
}
else
p->deletecount = 1;
p->deletecost = cost;
}
}
/* Perform insert-lines and delete-lines operations on CURRENT_MATRIX
according to the costs in MATRIX, using the direct scrolling method
which is used when the terminal supports setting a scroll window
(scroll_region_ok).
WINDOW_SIZE is the number of lines being considered for scrolling
and UNCHANGED_AT_TOP is the vpos of the first line being
considered. These two arguments can specify any contiguous range
of lines.
In the direct scrolling method, a new scroll window is selected
before each insertion or deletion, so that groups of lines can be
scrolled directly to their final vertical positions. This method
is described in more detail in calculate_direct_scrolling, where
the cost matrix for this approach is constructed. */
static void
do_direct_scrolling (struct frame *frame, struct glyph_matrix *current_matrix,
struct matrix_elt *cost_matrix, int window_size,
int unchanged_at_top)
{
struct matrix_elt *p;
int i, j;
/* A queue of deletions and insertions to be performed. */
struct alt_queue { int count, pos, window; };
struct alt_queue *queue_start = (struct alt_queue *)
alloca (window_size * sizeof *queue_start);
struct alt_queue *queue = queue_start;
/* Set to 1 if a terminal window has been set with
set_terminal_window: */
int terminal_window_p = 0;
/* A nonzero value of write_follows indicates that a write has been
selected, allowing either an insert or a delete to be selected
next. When write_follows is zero, a delete cannot be selected
unless j < i, and an insert cannot be selected unless i < j.
This corresponds to a similar restriction (with the ordering
reversed) in calculate_direct_scrolling, which is intended to
ensure that lines marked as inserted will be blank. */
int write_follows_p = 1;
/* For each row in the new matrix what row of the old matrix it is. */
int *copy_from = (int *) alloca (window_size * sizeof (int));
/* Non-zero for each row in the new matrix that is retained from the
old matrix. Lines not retained are empty. */
char *retained_p = (char *) alloca (window_size * sizeof (char));
memset (retained_p, 0, window_size * sizeof (char));
/* Perform some sanity checks when GLYPH_DEBUG is on. */
CHECK_MATRIX (current_matrix);
/* We are working on the line range UNCHANGED_AT_TOP ...
UNCHANGED_AT_TOP + WINDOW_SIZE (not including) in CURRENT_MATRIX.
We step through lines in this range from the end to the start. I
is an index into new lines, j an index into old lines. The cost
matrix determines what to do for ranges of indices.
If i is decremented without also decrementing j, this corresponds
to inserting empty lines in the result. If j is decremented
without also decrementing i, this corresponds to omitting these
lines in the new rows, i.e. rows are deleted. */
i = j = window_size;
while (i > 0 || j > 0)
{
p = cost_matrix + i * (window_size + 1) + j;
if (p->insertcost < p->writecost
&& p->insertcost < p->deletecost
&& (write_follows_p || i < j))
{
/* Insert is cheaper than deleting or writing lines. Leave
a hole in the result display that will be filled with
empty lines when the queue is emptied. */
queue->count = 0;
queue->window = i;
queue->pos = i - p->insertcount;
++queue;
i -= p->insertcount;
write_follows_p = 0;
}
else if (p->deletecost < p->writecost
&& (write_follows_p || i > j))
{
/* Deleting lines is cheaper. By decrementing J, omit
deletecount lines from the original. */
write_follows_p = 0;
j -= p->deletecount;
}
else
{
/* One or more lines should be written. In the direct
scrolling method we do this by scrolling the lines to the
place they belong. */
int n_to_write = p->writecount;
write_follows_p = 1;
xassert (n_to_write > 0);
if (i > j)
{
/* Immediately insert lines */
set_terminal_window (frame, i + unchanged_at_top);
terminal_window_p = 1;
ins_del_lines (frame, j - n_to_write + unchanged_at_top, i - j);
}
else if (i < j)
{
/* Queue the deletion of a group of lines */
queue->pos = i - n_to_write + unchanged_at_top;
queue->window = j + unchanged_at_top;
queue->count = i - j;
++queue;
}
while (n_to_write > 0)
{
--i, --j, --n_to_write;
copy_from[i] = j;
retained_p[j] = 1;
}
}
}
/* Do queued operations. */
if (queue > queue_start)
{
int next = -1;
do
{
--queue;
if (queue->count)
{
set_terminal_window (frame, queue->window);
terminal_window_p = 1;
ins_del_lines (frame, queue->pos, queue->count);
}
else
{
for (j = queue->window - 1; j >= queue->pos; --j)
{
while (retained_p[++next])
;
copy_from[j] = next;
}
}
}
while (queue > queue_start);
}
/* Now, for each row I in the range of rows we are working on,
copy_from[i] gives the original line to copy to I, and
retained_p[copy_from[i]] is zero if line I in the new display is
empty. */
mirrored_line_dance (current_matrix, unchanged_at_top, window_size,
copy_from, retained_p);
if (terminal_window_p)
set_terminal_window (frame, 0);
}
void
scrolling_1 (FRAME_PTR frame, int window_size, int unchanged_at_top,
int unchanged_at_bottom, int *draw_cost, int *old_draw_cost,
int *old_hash, int *new_hash, int free_at_end)
{
struct matrix_elt *matrix;
matrix = ((struct matrix_elt *)
alloca ((window_size + 1) * (window_size + 1) * sizeof *matrix));
if (FRAME_SCROLL_REGION_OK (frame))
{
calculate_direct_scrolling (frame, matrix, window_size,
unchanged_at_bottom,
draw_cost, old_draw_cost,
old_hash, new_hash, free_at_end);
do_direct_scrolling (frame, frame->current_matrix,
matrix, window_size, unchanged_at_top);
}
else
{
calculate_scrolling (frame, matrix, window_size, unchanged_at_bottom,
draw_cost, old_hash, new_hash,
free_at_end);
do_scrolling (frame,
frame->current_matrix, matrix, window_size,
unchanged_at_top);
}
}
/* Return number of lines in common between current and desired frame
contents described to us only as vectors of hash codes OLDHASH and
NEWHASH. Consider only vpos range START to END (not including
END). Ignore short lines on the assumption that avoiding redrawing
such a line will have little weight. */
int
scrolling_max_lines_saved (int start, int end,
int *oldhash, int *newhash,
int *cost)
{
struct { int hash; int count; } lines[01000];
register int i, h;
register int matchcount = 0;
int avg_length = 0;
int threshold;
/* Compute a threshold which is 1/4 of average length of these lines. */
for (i = start; i < end; i++)
avg_length += cost[i];
avg_length /= end - start;
threshold = avg_length / 4;
memset (lines, 0, sizeof lines);
/* Put new lines' hash codes in hash table. Ignore lines shorter
than the threshold. Thus, if the lines that are in common are
mainly the ones that are short, they won't count. */
for (i = start; i < end; i++)
{
if (cost[i] > threshold)
{
h = newhash[i] & 0777;
lines[h].hash = newhash[i];
lines[h].count++;
}
}
/* Look up old line hash codes in the hash table. Count number of
matches between old lines and new. */
for (i = start; i < end; i++)
{
h = oldhash[i] & 0777;
if (oldhash[i] == lines[h].hash)
{
matchcount++;
if (--lines[h].count == 0)
lines[h].hash = 0;
}
}
return matchcount;
}
/* Calculate the line insertion/deletion
overhead and multiply factor values */
static void
line_ins_del (FRAME_PTR frame, int ov1, int pf1, int ovn, int pfn,
register int *ov, register int *mf)
{
register int i;
register int frame_lines = FRAME_LINES (frame);
register int insert_overhead = ov1 * 10;
register int next_insert_cost = ovn * 10;
for (i = frame_lines-1; i >= 0; i--)
{
mf[i] = next_insert_cost / 10;
next_insert_cost += pfn;
ov[i] = (insert_overhead + next_insert_cost) / 10;
insert_overhead += pf1;
}
}
static void
ins_del_costs (FRAME_PTR frame,
const char *one_line_string, const char *multi_string,
const char *setup_string, const char *cleanup_string,
int *costvec, int *ncostvec,
int coefficient)
{
if (multi_string)
line_ins_del (frame,
string_cost (multi_string) * coefficient,
per_line_cost (multi_string) * coefficient,
0, 0, costvec, ncostvec);
else if (one_line_string)
line_ins_del (frame,
string_cost (setup_string) + string_cost (cleanup_string), 0,
string_cost (one_line_string),
per_line_cost (one_line_string),
costvec, ncostvec);
else
line_ins_del (frame,
9999, 0, 9999, 0,
costvec, ncostvec);
}
/* Calculate the insert and delete line costs.
Note that this is done even when running with a window system
because we want to know how long scrolling takes (and avoid it).
This must be redone whenever the frame height changes.
We keep the ID costs in a precomputed array based on the position
at which the I or D is performed. Also, there are two kinds of ID
costs: the "once-only" and the "repeated". This is to handle both
those terminals that are able to insert N lines at a time (once-
only) and those that must repeatedly insert one line.
The cost to insert N lines at line L is
[tt.t_ILov + (frame_lines + 1 - L) * tt.t_ILpf] +
N * [tt.t_ILnov + (frame_lines + 1 - L) * tt.t_ILnpf]
ILov represents the basic insert line overhead. ILpf is the padding
required to allow the terminal time to move a line: insertion at line
L changes (frame_lines + 1 - L) lines.
The first bracketed expression above is the overhead; the second is
the multiply factor. Both are dependent only on the position at
which the insert is performed. We store the overhead in
FRAME_INSERT_COST (frame) and the multiply factor in
FRAME_INSERTN_COST (frame). Note however that any insertion
must include at least one multiply factor. Rather than compute this
as FRAME_INSERT_COST (frame)[line]+FRAME_INSERTN_COST (frame)[line],
we add FRAME_INSERTN_COST (frame) into FRAME_INSERT_COST (frame).
This is reasonable because of the particular algorithm used in calcM.
Deletion is essentially the same as insertion.
*/
void
do_line_insertion_deletion_costs (FRAME_PTR frame,
const char *ins_line_string,
const char *multi_ins_string,
const char *del_line_string,
const char *multi_del_string,
const char *setup_string,
const char *cleanup_string,
int coefficient)
{
FRAME_INSERT_COST (frame) =
xnrealloc (FRAME_INSERT_COST (frame), FRAME_LINES (frame), sizeof (int));
FRAME_DELETEN_COST (frame) =
xnrealloc (FRAME_DELETEN_COST (frame), FRAME_LINES (frame), sizeof (int));
FRAME_INSERTN_COST (frame) =
xnrealloc (FRAME_INSERTN_COST (frame), FRAME_LINES (frame), sizeof (int));
FRAME_DELETE_COST (frame) =
xnrealloc (FRAME_DELETE_COST (frame), FRAME_LINES (frame), sizeof (int));
ins_del_costs (frame,
ins_line_string, multi_ins_string,
setup_string, cleanup_string,
FRAME_INSERT_COST (frame), FRAME_INSERTN_COST (frame),
coefficient);
ins_del_costs (frame,
del_line_string, multi_del_string,
setup_string, cleanup_string,
FRAME_DELETE_COST (frame), FRAME_DELETEN_COST (frame),
coefficient);
}
Reference in a new issue View git blame Copy permalink