Use them instead of malloc, xmalloc, and xfree, for Lisp objects.
Don't set allocating_for_lisp in the callers; let lisp_malloc do it.
(n_interval_blocks, n_float_blocks): New variable.
(n_cons_blocks, n_vectors, n_symbol_blocks): New variable.
(n_marker_blocks, n_string_blocks): New variable.
(init_intervals, make_interval): Set a count variable.
Use lisp_malloc instead of setting allocating_for_lisp.
(init_float, make_float, init_cons, Fcons): Likewise.
(allocate_vectorlike, init_symbol, Fmake_symbol): Likewise
(init_marker, allocate_misc, init_strings): Likewise.
(make_uninit_multibyte_string): Likewise.
(gc_sweep, compact_strings): Decrement the count variables.
(uninterrupt_malloc): Don't store Emacs's hooks
into the old_..._hook variables.
by complete_deferred_msg.
(post_character_message): Cancel all deferred messages when
quit_char is typed, in order to break out of potential deadlocks.
(cancel_all_deferred_msgs): New function.
(complete_deferred_msg): Don't abort if msg not found; may have
been cancelled.
(Fw32_reconstruct_hot_key): Use pre-interned symbols.
(Fw32_send_sys_command): Wait for system command to
complete before returning.
Do free malloc'd memory even in case of failure.
Use xfree, not free.
(Fbase64_encode_region, Fbase64_encode_string):
Use xfree, not free.
(base64_decode_1): Don't fail for short lines.
(Fcall_process): Doc-string modified. Set CODING_MODE_LAST_BLOCK
bit in argument_coding.mode to tell encoding procedure to flush
data. While encoding arguments, if argument_coding uses CCL
program, initialize it again after encoding.
(Fcall_process_region): Doc-string modified. Handle the case that
NARGS is 3.
(ccl_driver) <CCL_End>: Pretend that all source text is consumed.
Set `ic' to the head of this command.
(ccl_driver) <CCL_ReadMultibyteChar2>: Return a single byte 8-bit
code for an invalid code.
(setup_ccl_program): If the arg VEC is nil, skip setting ups
based on VEC.
(init_charset_once): Intern and staticpro Qunknown. Initialize
all elements of Vcharset_symbol_table to Qunknown.
(find_charset_in_str): New arg MULTIBYTE. If it is zero, check
unibyte characters only. For an invalid composition sequence, set
CHARSETS[1] to 1.
(Ffind_charset_region): Call find_charset_in_str with an
appropriate MULTIBYTE arg. If undefined charsets are found,
include `unknown' is the return value.
(Ffind_charset_string): Likewise.
(Fsplit_char): If CHAR is invalid, return `(unknown CHAR)'.
(str_cmpchar_id): Max composite character code should be less than
GENERIC_COMPOSITION_CHAR.
final_char is invalid.
(decode_coding): If coding->type is coding_type_ccl, call
ccl_coding_driver even if SRC_BYTES is zero.
(code_convert_region): Update `inserted' correctly after calling
coding->post_read_conversion. Even after the code converter
consumed all source text, call it once more if it is
coding_type_ccl to flush out data.
CODING_MODE_LAST_BLOCK bit is not yet set in coding->mode, set it
and try decoding again.
(Fprocess_send_eof): If the coding system of PROCESS
requires flushing, call send_process with null-string at first.
to search for an invalid multibyte code.
In unibyte case, no need to check whether there are translations
in more than one charset; just set charset_base to 0.
(syms_of_msdos): DEFVAR_LISP it.
(IT_insert_glyphs, IT_delete_glyphs): New functions which abort
Emacs.
(internal_terminal_init): Set up insert_glyphs_hook and
delete_glyphs_hook to call them. Explicitly set char_ins_del_ok
to 0.
(unibyte_display_via_language_environment): New variable.
(syms_of_msdos): Devfar it.
(IT_write_glyphs): Honor glyph aliasing via Vglyph_table.
Encode the character codes of the glyphs according to the
terminal_coding in effect.
