coding->dst_multibyte before using CODING_REQUIRE_DECODING.
(encode_coding_string): Set coding->src_multibyte and
coding->dst_multibyte before using CODING_REQUIRE_ENCODING.
(conversion_buffer, conversion_buffer_size): Variables deleted.
(get_conversion_buffer): Function deleted.
(struct conversion_buffer): New structure.
(MAX_ALLOCA): New macro.
(allocate_conversion_buffer): New macro.
(extend_conversion_buffer, free_conversion_buffer): New functions.
(ccl_coding_driver): Set coding->result.
(decode_coding): Set coding->result to CODING_FINISH_NORMAL if
this is the last block of source.
(encode_coding): Likewise. Handle the source block as the last
one only when the whole source text is consumed.
(decode_coding_string): Handle the case that the output buffer is
too small to decode the whole source text. Use
allocate_conversion_buffer, extend_conversion_buffer and
free_conversion_buffer, not get_conversion_buffer.
(encode_coding_string): Likewise.
(init_coding): Function deleted.
(init_coding_once): Delete code to initialize
conversion_buffer_size.
cmp_data is not set properly (because of invalid code in the
source text).
(run_pre_post_conversion_on_str): Include text properties in the
resulting string.
(decode_coding_string): Set members of coding correctly.
(Qsafe_chars, Vchar_coding_system_table, Qchar_coding_system): New
variables.
(coding_safe_chars): New function.
(CODING_SAFE_CHAR_P): New macro.
(CHARSET_OK): New arg C. Call CODING_SAFE_CHAR_P instead of
checking safe_charsets member of the coding system. Caller
changed.
(detect_coding_iso2022): New local variable safe_chars.
(DECODE_DESIGNATION): Call CODING_SAFE_CHAR_P instead of checking
safe_charsets member of the coding system.
(decode_coding_iso2022): New local variable safe_chars.
(ENCODE_ISO_CHARACTER_DIMENSION1): Don't check unsafe chars here.
(ENCODE_ISO_CHARACTER_DIMENSION2): Likewise.
(ENCODE_ISO_CHARACTER): Arguments changed. Caller changed.
(ENCODE_UNSAFE_CHARACTER): New macro.
(encode_coding_iso2022): New local variable safe_chars. Check
unsafe chars.
(setup_coding_system): Delete the code to initialize
coding->safe_charses
(intersection, find_safe_codings): New functions.
(Ffind_coding_systems_region_internal): New function.
(syms_of_coding): Defsubr it. Initialize Qsafe_chars,
Qsafe_cding_system. Make Vchar_coding_system_table a Lisp
variable and initialize it.
(ccl_coding_driver): On encoding, initialize ccl->eol_type.
(decode_eol_post_ccl): New function.
(decode_coding): Don't detect EOL format here for CCL based coding
systems.
(decode_coding) <coding_type_ccl>: Handle carryovered CR. Call
decode_eol_post_ccl after running the CCL program.
(code_convert_region): Don't detect EOL format here for CCL based
coding systems.
(decode_coding_string): Likewise.
yet allocated, finish decoding with result
CODING_FINISH_INSUFFICIENT_CMP.
(coding_allocate_composition_data): Make it non-static.
(coding_restore_composition): Likewise.
beginning of buffer before calling coding->post_read_conversion.
(decode_coding_string): Give correct args to
run_pre_post_conversion_on_str.
(encode_coding_string): Likewise.
CODING_FINISH_INSUFFICIENT_SRC if there's not enough source.
(ONE_MORE_CHAR, EMIT_CHAR, EMIT_ONE_BYTE, EMIT_TWO_BYTE,
EMIT_BYTES): New macros.
(THREE_MORE_BYTES, DECODE_CHARACTER_ASCII,
DECODE_CHARACTER_DIMENSION1, DECODE_CHARACTER_DIMENSION2): These
macros deleted.
(CHECK_CODE_RANGE_A0_FF): This macro deleted.
(detect_coding_emacs_mule): Use UNIBYTE_STR_AS_MULTIBYTE_P to
check the validity of multibyte sequence.
(decode_coding_emacs_mule): New function.
(encode_coding_emacs_mule): New macro.
(detect_coding_iso2022): Use ONE_MORE_BYTE to fetch a byte from
the source.
(DECODE_ISO_CHARACTER): Just return a character code.
(DECODE_COMPOSITION_START): Set coding->result instead of result.
(decode_coding_iso2022, decode_coding_sjis_big5, decode_eol): Use
EMIT_CHAR to produced decoded characters. Exit the loop only by
macros ONE_MORE_BYTE or EMIT_CHAR. Don't handle the case of last
block here.
(ENCODE_ISO_CHARACTER): Don't translate character here. Produce
only position codes for an invalid character.
(encode_designation_at_bol): Return new destination pointer. 5th
arg DSTP is changed to DST.
(encode_coding_iso2022, decode_coding_sjis_big5): Get a character
from the source by ONE_MORE_CHAR. Don't handle the case of last
block here.
(DECODE_SJIS_BIG5_CHARACTER, ENCODE_SJIS_BIG5_CHARACTER): These
macros deleted.
(detect_coding_sjis, detect_coding_big5, detect_coding_utf_8,
detect_coding_utf_16, detect_coding_ccl): Use ONE_MORE_BYTE and
TWO_MORE_BYTES to fetch a byte from the source.
(encode_eol): Pay attention to coding->src_multibyte.
(detect_coding, detect_eol): Preserve members src_multibyte and
dst_multibyte.
(DECODING_BUFFER_MAG): Return 2 even for coding_type_raw_text.
(encoding_buffer_size): Set magnification to 3 for all coding
systems that require encoding.
(ccl_coding_driver): For decoding, be sure that the result is
valid multibyte sequence.
(decode_coding): Initialize coding->errors and coding->result.
For emacs-mule, call decode_coding_emacs_mule. For no-conversion
and raw-text, always call decode_eol. Handle the case of last
block here. If not coding->dst_multibyte, convert the resulting
sequence to unibyte.
(encode_coding): Initialize coding->errors and coding->result.
For emacs-mule, call encode_coding_emacs_mule. For no-conversion
and raw-text, always call encode_eol. Handle the case of last
block here.
(shrink_decoding_region, shrink_encoding_region): Detect cases
that we can't skip data more rigidly.
(code_convert_region): Setup src_multibyte and dst_multibyte
members of coding. For decoding, if the buffer is multibyte,
convert the source sequence to unibyte in advance. For encoding,
if the buffer is multibyte, convert the resulting sequence to
multibyte afterward.
(run_pre_post_conversion_on_str): New function.
(code_convert_string): Deleted and divided into the following two.
(decode_coding_string, encode_coding_string): New functions.
(code_convert_string1, code_convert_string_norecord): Call one of
above.
(Fdecode_sjis_char, Fdecode_big5_char): Use MAKE_CHAR instead of
MAKE_NON_ASCII_CHAR.
(Fset_terminal_coding_system_internal,
Fset_safe_terminal_coding_system_internal): Setup src_multibyte
and dst_multibyte members.
(init_coding_once): Initialize iso_code_class with new enum
ISO_control_0 and ISO_control_1.
coding-category-utf-16-be, and coding-category-utf-16-le.
(coding_category_name): Include "coding-category-utf-8",
"coding-category-utf-16-be", and "coding-category-utf-16-le".
(UTF_8_1_OCTET_P) (UTF_8_EXTRA_OCTET_P) (UTF_8_2_OCTET_LEADING_P)
(UTF_8_3_OCTET_LEADING_P) (UTF_8_4_OCTET_LEADING_P)
(UTF_8_5_OCTET_LEADING_P) (UTF_8_6_OCTET_LEADING_P): New macros.
(detect_coding_utf_8): New function.
(UTF_16_INVALID_P) (TF_16_HIGH_SURROGATE_P)
(UTF_16_LOW_SURROGATE_P): New macros.
(detect_coding_utf_16): New function.
(detect_coding_mask): Fix bug of returning wrong mask bits in the
case that detect_coding_XXX returns a mask not set in
priorities[i].
(detect_eol_type_in_2_octet_form): New function.
(detect_eol): If cooding->category_idx is for UTF-16, call
detect_eol_type_in_2_octet_form instead of dectect_eol_type.
(detect_coding_system): Don't include `nil' coding-system in the
result.
(Fupdate_coding_systems_internal): Update all coding-categories.
(DECODE_CHARACTER_ASCII): Don't handle composition here.
(DECODE_CHARACTER_DIMENSION1): Likewise. Don't check the validity
of multibyte code here.
(DECODE_CHARACTER_DIMENSION2): Likewise.
(detect_coding_emacs_mule): Change the case label from
EMACS_leading_code_composition to 0x80.
(detect_coding_iso2022): Handle new composition sequence.
(DECODE_ISO_CHARACTER): Likewise.
(check_composing_code): Deleted.
(coding_allocate_composition_data): New function.
(CODING_ADD_COMPOSITION_START) (CODING_ADD_COMPOSITION_END)
(CODING_ADD_COMPOSITION_COMPONENT) (DECODE_COMPOSITION_START)
(DECODE_COMPOSITION_END) (DECODE_COMPOSITION_RULE): New macros.
(decode_coding_iso2022): Handle new composition sequence.
(ENCODE_ISO_CHARACTER): Don't check composition here.
(ENCODE_COMPOSITION_RULE) (ENCODE_COMPOSITION_START): New macros.
(ENCODE_COMPOSITION_NO_RULE_START)
(ENCODE_COMPOSITION_WITH_RULE_START): Deleted.
(ENCODE_COMPOSITION_END): Handle new composition sequence.
(ENCODE_COMPOSITION_FAKE_START): New macro.
(encode_coding_iso2022): Handle new composition sequence.
(ENCODE_SJIS_BIG5_CHARACTER): Delete superfluous `;' at the tail.
(encode_coding_sjis_big5): Ignore composition.
(setup_coding_system): Initialize new members of struct
coding_system. Enable composition only when the coding system has
`composition' property t.
(coding_free_composition_data) (coding_adjust_composition_offset)
(coding_save_composition) (coding_restore_composition): New
functions.
(code_convert_region): Call coding_save_composition for encoding
and coding_allocate_composition_data for decoding. Don't skip
ASCII characters if we handle composition on encoding. Call
signal_after_change with Check_BORDER.
(code_convert_string): Call coding_save_composition for encoding
and coding_allocate_composition_data for decoding. Don't skip
ASCII characters if we handle composition on encoding.
(code_convert_string1): Set Vlast_coding_system_used after calling
code_convert_string.
(code_convert_string_norecord): Disable composition.
(Fset_terminal_coding_system_internal): Likewise.
(Fset_safe_terminal_coding_system_internal): Likewise.
(Fset_keyboard_coding_system_internal): Likewise.
(init_coding_once): Set emacs_code_class[0x80] to
EMACS_invalid_code.
* fileio.c (strerror): Likewise.
* process.c (strerror): Likewise.
* emacs.c (strerror): Likewise.
(Vsystem_messages_locale): Renamed from Vmessages_locale.
All uses changed.
(Vprevious_system_messages_locale): Likewise, from
Vprevious_messages_locale.
(Vsystem_time_locale): Likewise, from Vtime_locale.
(Vprevious_system_time_locale): Likewise, from Vprevious_time_locale.
(ABORT_RETURN_TYPE): New macro.
(abort): Return type is now ABORT_RETURN_TYPE.
(main): Always invoke init_signals, even if POSIX_SIGNALS is not
defined.
(syms_of_emacs): messages-locale -> system-messages-locale,
previous-messages-locale -> previous-system-messages-locale,
time-locale -> system-time-locale,
previous-time-locale -> previous-system-time-locale.
* gmalloc.c (PP, __ptr_t): Assume ANSI C if STDC_HEADERS is defined.
(const): Do not define; that's config.h's job.
(<limits.h>): Include if HAVE_LIMITS_H is defined.
(CHAR_BIT): Move test for definedness outside of limits.h condition.
(<stddef.h>): Include if STDC_HEADERS is defined.
(FREE_RETURN_TYPE): New macro.
(free): Return type is now FREE_RETURN_TYPE.
* lisp.h (synchronize_system_time_locale): Renamed from
synchronize_time_locale. All uses changed.
(synchronize_system_messages_locale): Likewise, from
synchronize_messages_locale.
* process.c (sys_siglist): Remove.
* syntax.c (scan_sexps_forward): Use abort, not assert.
* sysdep.c (my_sys_siglist): New var.
(sys_siglist): New macro. Remove old initialized vars of same name.
(init_signals): Initialize sys_siglist.
* xfns.c (abort): Remove decl; stdlib.h now does this.
* Makefile.in (emacs): Set the LC_ALL environment variable to "C"
when dumping, so that the dumped Emacs doesn't have stray locale info.
(dired.o): Depend on systime.h.
(editfns.o): Depend on coding.h.
* alloc.c, buffer.c, callproc.c, ccl.c, charset.c, coding.c, data.c,
dispnew.c, editfns.c, emacs.c, filelock.c, floatfns.c, hftctl.c,
keyboard.c, process.c, sysdep.c, unexelf.c, unexhp9k800.c,
unexsunos4.c, vmsfns.c, vmsgmalloc.c, w32faces.c, w32menu.c, w32term.c,
w32xfns.c, xfaces.c, xfns.c, xmenu.c, xterm.c:
Include <config.h> before any system include files.
* alloc.c, buffer.c, ccl.c, data.c, editfns.c, emacs.c, eval.c,
fileio.c, filelock.c, frame.c, insdel.c, keymap.c, lread.c,
m/alpha.h, print.c, search.c, sysdep.c, xdisp.c, xfaces.c, xfns.c,
xmenu.c, xterm.c:
Do not include <stdlib.h>, as <config.h> does this now.
* callproc.c (Fcall_process):
Synchronize messages locale before invoking strerror.
Decode resulting string with locale-coding-system.
* coding.c (Vlocale_coding_system): New var.
(syms_of_coding): Adjust to above change.
(emacs_strerror): New function.
* coding.h (emacs_strerror, Vlocale_coding_system): New decls.
* config.in (HAVE_STDIO_EXT_H, HAVE_TM_GMTOFF, HAVE___FPENDING,
HAVE_FTELLO, HAVE_GETLOADAVG, HAVE_MBLEN, HAVE_MBRLEN,
HAVE_STRSIGNAL): New macros.
(BITS_PER_LONG): Default to 64 if _LP64 is defined.
<stdlib.h>: Include if HAVE_STDLIB_H is defined and NOT_C_CODE isn't.
* dired.c: Include "systime.h".
(Ffile_attributes): Do not cast s.st_size to int; this loses
information if int is 32 bits but st_size and EMACS_INT are larger.
Treat large device numbers like large inode numbers.
* dispnew.c (PENDING_OUTPUT_COUNT): Use __fpending if available.
* editfns.c: Include coding.h.
(emacs_strftime): Remove decl.
(emacs_strftimeu): New decl.
(emacs_memftimeu): Renamed from emacs_memftime; new arg UT.
Use emacs_strftimeu instead of emacs_strftime.
(Fformat_time_string): Convert format string using
Vlocale_coding_system, and convert result back. Synchronize time
locale before invoking lower level function. Invoke
emacs_memftimeu, passing ut, instead of emacs_memftime.
* emacs.c: Include <locale.h> if HAVE_SETLOCALE is defined.
(Vmessages_locale, Vprevious_messages_locale, Vtime_locale,
Vprevious_time_locale): New variables.
(main): Invoke setlocale early, so that initial error messages are
localized properly. But skip locale-setting if LC_ALL is "C".
Fix up locale when it's safe to do so.
(fixup_locale): Moved here from xterm.c.
(synchronize_locale, synchronize_time_locale,
synchronize_messages_locale): New functions.
(syms_of_emacs): Accommodate above changes.
* fileio.c (report_file_error): Convert strerror output according
to Vlocale_coding_system.
(Finsert_file_contents): Check for arithmetic overflow in
computations that depend on file size. Report IO errors
with emacs_strerror, not strerror.
* fns.c (Fgethash): Declare dflt parameter.
* gmalloc.c: Do not define const to nothing if HAVE_CONFIG_H
is defined; that's config.h's job.
* lisp.h (EMACS_INT, BITS_PER_EMACS_INT, EMACS_UINT): If _LP64,
default these values to long, BITS_PER_LONG, and unsigned long.
(VALBITS, MARKBIT, XINT): Do not assume 32-bit EMACS_INT.
(PNTR_COMPARISON_TYPE): Default to EMACS_UINT, not to unsigned int.
(code_convert_string_norecord, fixup_locale,
synchronize_messages_locale, synchronize_time_locale,
emacs_open, emacs_close, emacs_read, emacs_write): New decls.
All Emacs callers of open, close, read, write changed to use
emacs_open, emacs_close, emacs_read, emacs_write.
* lread.c (file_offset, file_tell): New macros. All uses of ftell
changed to file_tell.
(saved_doc_string_position, prev_saved_doc_string_position): Now
of type file_offset.
(init_lread): Do not fix locale here; fixup_locale now does this.
* m/amdahl.h, s/usg5-4.h:
(NSIG): Remove.
(NSIG_MINIMUM): New macro.
* m/cydra5.h, m/dpx2.h, m/mips.h, m/pfa50.h, m/sps7.h, m/stride.h,
m/ustation.h, s/gnu-linux.h, s/hpux.h, s/iris3-5.h, s/iris3-6.h,
s/umips.h, s/usg5-4.h:
(SIGIO): Do not undef.
(BROKEN_SIGIO): New macro.
* m/ustation.h:
(SIGTSTP): Do not undef.
(BROKEN_SIGTSTP): New macro.
* s/gnu-linux.h:
(SIGPOLL, SIGURG): Do not undef.
(BROKEN_SIGPOLL, BROKEN_SIGURG): New macros.
* s/ptx4.h:
(SIGINFO): Do not undef.
(BROKEN_SIGINFO): New macros.
* m/delta.h, s/ptx.h, s/template.h: Doc fix.
* mktime.c, strftime.c: Update to glibc 2.1.2 version, with
some Emacs-related changes merged.
* print.c (float_to_string): Prepend "-" to representation of a
NaN if the NaN is negative.
* process.c (sys_siglist): Omit if HAVE_STRSIGNAL.
(wait_reading_process_input): Use emacs_strerror, not strerror.
* process.c (status_message, sigchld_handler): Synchronize locale,
then use strsignal istead of sys_siglist.
* w32proc.c (sys_wait): Likewise.
* s/aix3-1.h, s/bsd4-1.h, s/dgux.h, s/gnu-linux.h, s/hiuxmpp.h,
s/hpux.h, s/iris3-5.h, s/iris3-6.h, s/irix3-3.h, s/osf1.h, s/rtu.h,
s/sunos4-1.h, s/unipl5-0.h, s/unipl5-2.h, s/usg5-0.h, s/usg5-2-2.h,
s/usg5-2.h, s/usg5-3.h, s/xenix.h:
(open, close, read, write, INTERRUPTIBLE_OPEN,
INTERRUPTIBLE_CLOSE, INTERRUPTIBLE_IO): Remove.
* s/sol2-5.h (_LARGEFILE_SOURCE, _FILE_OFFSET_BITS): New macros.
* sysdep.c (sys_read, sys_write, read, write, sys_close, close,
sys_open, open): Remove.
(emacs_open, emacs_close, emacs_read, emacs_write): Always define;
the old INTERRUPTIBLE_OPEN, INTERRUPTIBLE_CLOSE, and INTERRUPTIBLE_IO
macros are no longer used.
(emacs_open): Renamed from sys_open. Merge BSD4_1 version.
(emacs_close): Renamed from sys_close.
(emacs_read): Renamed from sys_read.
(emacs_write): Renamed from sys_write.
(sys_siglist): Do not declare if HAVE_STRSIGNAL.
(dup2): Do not print error on failure; the real dup2 doesn't.
(strsignal): New function, defined if !HAVE_STRSIGNAL.
* syssignal.h (SIGINFO): Undef if defined and if BROKEN_SIGINFO
is defined.
(SIGIO, SIGPOLL, SIGTSTP, SIGURG): Likewise.
(NSIG): If less than NSIG_MINIMUM, define to NSIG_MINIMUM.
(strsignal): Declare if !HAVE_STRSIGNAL.
* unexelf.c (ElfBitsW, ELFSIZE, ElfExpandBitsW): New macros.
(ElfW): Define in terms of ElfExpandBitsW.
* w32proc.c (sys_siglist): Remove decl.
* xdisp.c (decode_mode_spec): 3rd arg is int, not char, to comply
with ANSI C.
(display_string): Declare face_string_pos arg.
* xfns.c (Fx_show_tip): Declare timeout param.
* xterm.c: No need to include locale.h.
(x_alloc_lighter_color, x_setup_relief_color):
Pass arg as double, not float, for compatibility with ANSI C.
(fixup_locale): Move to emacs.c.
(x_term_init): Do not setlocale or fixup locale; the main program
does this now.
