[COMMIT] Drop Ben's attempt to make Big5 into one charset, unicode-internal.

APPROVE COMMIT NOTE: This patch has been committed # HG changeset patch # User Aidan Kehoe <kehoea(a)parhasard.net&gt; # Date 1507320562 -3600 # Fri Oct 06 21:09:22 2017 +0100 # Node ID e3925877abce049cee4c83fcfd8a78a875ee813b # Parent 19876e056bb7630f1854719c8b037a1f2fb691e8 Drop Ben's attempt to make Big5 into one charset, unicode-internal. src/ChangeLog addition: 2017-10-06 Aidan Kehoe <kehoea(a)parhasard.net&gt; * lisp.h: * mule-charset.c: * mule-charset.c (syms_of_mule_charset): * mule-charset.c (complex_vars_of_mule_charset): * mule-charset.c (init_mule_charset): * mule-coding.c: * mule-coding.c (byte_big5_two_byte_1_p): * mule-coding.c (big5_decode): * mule-coding.c (big5_encode): * mule-coding.c (decode_big5_char): * mule-coding.c (big5_char_to_fake_codepoint): * mule-coding.c (Fencode_big5_char): * mule-coding.c (init_mule_coding): * text.c: * unicode.c (verify_load_unicode_args): Remove the attempt to implement Big5 as one charset under unicode-internal; this makes unicode-internal builds incompatible with byte-code compiled by 21.4 and even source code, e.g. leim/quail/Punct-B5.el. lisp/ChangeLog addition: 2017-10-06 Aidan Kehoe <kehoea(a)parhasard.net&gt; * mule/chinese.el: * mule/chinese.el (for): * mule/chinese.el (chinese-big5/list): * mule/chinese.el (find-charset): * mule/mule-category.el (predefined-category-list): * mule/mule-charset.el: Remove Ben's attempt under unicode-internal to make Big5 into one charset. This is incompatible with byte-compiled code from 21.4, which uses the hackish chinese-big5-1, chinese-big5-2 charsets with ISO-2022 escapes to encode these characters. If at some point in the future where we use UTF-8 for compiled code this could be revisited. diff -r 19876e056bb7 -r e3925877abce lisp/ChangeLog --- a/lisp/ChangeLog Thu Oct 05 00:08:17 2017 +0100 +++ b/lisp/ChangeLog Fri Oct 06 21:09:22 2017 +0100 ＠＠ -1,3 +1,18 ＠＠ +2017-10-06 Aidan Kehoe <kehoea(a)parhasard.net&gt; + + * mule/chinese.el: + * mule/chinese.el (for): + * mule/chinese.el (chinese-big5/list): + * mule/chinese.el (find-charset): + * mule/mule-category.el (predefined-category-list): + * mule/mule-charset.el: + Remove Ben's attempt under unicode-internal to make Big5 into one + charset. This is incompatible with byte-compiled code from 21.4, + which uses the hackish chinese-big5-1, chinese-big5-2 charsets + with ISO-2022 escapes to encode these characters. + If at some point in the future where we use UTF-8 for compiled + code this could be revisited. + 2017-09-30 Aidan Kehoe <kehoea(a)parhasard.net&gt; * mule/thai.el: diff -r 19876e056bb7 -r e3925877abce lisp/mule/chinese.el --- a/lisp/mule/chinese.el Thu Oct 05 00:08:17 2017 +0100 +++ b/lisp/mule/chinese.el Fri Oct 06 21:09:22 2017 +0100 ＠＠ -37,11 +37,8 ＠＠ ;; Setup auto-fill-chars for charsets that should invoke auto-filling. ;; SPACE and NEWLINE are already set. -(loop for l in `(chinese-gb2312 - ,＠(if (find-charset 'chinese-big5-1) - '(chinese-big5-1 chinese-big5-2) - '(chinese-big5))) - do (put-char-table l t auto-fill-chars)) +(loop for l in '(chinese-gb2312 chinese-big5-1 chinese-big5-2) + do (put-char-table l t auto-fill-chars)) ;; CNS11643 Plane3 thru Plane7 ;; These represent more and more obscure Chinese characters. ＠＠ -111,11 +108,9 ＠＠ :list '(chinese-cns11643-1 chinese-cns11643-2 chinese-cns11643-3 chinese-cns11643-4 chinese-cns11643-5 chinese-cns11643-6 chinese-cns11643-7)) -(define-charset-tag 'chinese-big5/list - :list (if (featurep 'unicode-internal) '(chinese-big5) - '(chinese-big5-1 chinese-big5-2))) -(define-charset-tag 'chinese-gb/list - :list '(chinese-gb2312 chinese-sisheng)) +(define-charset-tag 'chinese-big5/list :list '(chinese-big5-1 chinese-big5-2)) + +(define-charset-tag 'chinese-gb/list :list '(chinese-gb2312 chinese-sisheng)) (define-charset-tag 'chinese-gb-env/list :list '(chinese-gb/list chinese-big5/list chinese-cns/list chinese/language)) ＠＠ -302,13 +297,8 ＠＠ (define-coding-system-alias 'cn-big5 'big5) -;; Need to use `compiled-when' because otherwise we will get an error when -;; compiling this file under Unicode-internal because (charset-id -;; 'chinese-big5-2) gets evaluated at compile time and fails. -(compiled-when (find-charset 'chinese-big5-1) - ;; Big5 font requires special encoding. But under Unicode-internal we - ;; have one single charset `chinese-big5', with no special encoding - ;; needed. +(when (find-charset 'chinese-big5-1) + ;; Big5 font requires special encoding. (define-ccl-program ccl-encode-big5-font `(0 ;; In: R0:chinese-big5-1 or chinese-big5-2 diff -r 19876e056bb7 -r e3925877abce lisp/mule/mule-category.el --- a/lisp/mule/mule-category.el Thu Oct 05 00:08:17 2017 +0100 +++ b/lisp/mule/mule-category.el Fri Oct 06 21:09:22 2017 +0100 ＠＠ -238,10 +238,8 ＠＠ ;;much better category definitions. ;;(chinese-cns11643-1 ?t "Chinese Taiwan (CNS or Big5) 2-byte character set") (chinese-cns11643-2 ?t "Chinese Taiwan (CNS or Big5) 2-byte character set") - ,＠(if (find-charset 'chinese-big5-1) - '((chinese-big5-1 ?t) - (chinese-big5-2 ?t)) - '((chinese-big5 ?t))) + (chinese-big5-1 ?t) + (chinese-big5-2 ?t) (korean-ksc5601 ?h "Hangul (Korean) 2-byte character set") (jit-ucs-charset-0 ?J "Just-in-time-allocated Unicode character") ) diff -r 19876e056bb7 -r e3925877abce lisp/mule/mule-charset.el --- a/lisp/mule/mule-charset.el Thu Oct 05 00:08:17 2017 +0100 +++ b/lisp/mule/mule-charset.el Fri Oct 06 21:09:22 2017 +0100 ＠＠ -1254,73 +1254,64 ＠＠ tags (cns kanji traditional-chinese chinese/language) )) -(if (featurep 'unicode-internal) - ;; We can support Big5 directly. - (make-charset - 'chinese-big5 - "Big5 (Chinese traditional)" - '(dimension 2 - ;; Big5 claims to be a 94x157 charset, but with gaps in the middle. - ;; In particular, the rows are (theoretically) indexed from A1 - FE - ;; and the columns from 40 - 7E and A1 - FE. In fact, there are gaps - ;; in the rows as well (rows C7 and C8 are missing, as well as rows - ;; FA - FE), but that appears to be due to accident -- i.e. they just - ;; ran out of chars and/or wanted to make room for expansion. Note - ;; also that the gap at C7 and C8 is due to the Level-1/Level-2 - ;; division of Big5 (see below). The 94 rows are those between - ;; A1 and FE, inclusive. The 157 columns count the sum of the columns - ;; in each disjoint set. For us, we need to use the size of the range - ;; [40, FE], which is 191. - chars (94 191) - offset (161 64) - short-name "Chinese traditional (Big5)" - long-name "Chinese traditional (Big5)" - registries ["big5.eten-0"] - unicode-map ("unicode/unicode-consortium/EASTASIA/OBSOLETE/BIG5.TXT") - tags (kanji traditional-chinese chinese/language) - )) - ;; Old Mule situation; we can only handle up to 96x96 charsets. - ;; So we split it into two charsets. According to Ken Lunde's CJKV - ;; book, Big5 itself is split into "Big Five Level 1" (rows A1-C6) - ;; and "Big Five Level 2" (rows C9-F9), with the latter containing - ;; less used characters. We split the same way then coerce the - ;; result into a 94x94 block. - (make-charset - 'chinese-big5-1 - "Frequently used part (A141-C67F) of Big5 (Chinese traditional)" - '(dimension 2 - chars 94 - final ?0 - graphic 0 - short-name "Chinese traditional (Big5), L1" - long-name "Chinese traditional (Big5) (Level-1) A141-C67F" - registries ["big5.eten-0"] - ;; no unicode map, see chinese-big5-2 - tags (kanji traditional-chinese chinese/language) - )) - (make-charset - 'chinese-big5-2 - "Less frequently used part (C940-FEFE) of Big5 (Chinese traditional)" - '(dimension 2 - chars 94 - final ?1 - graphic 0 - short-name "Chinese traditional (Big5), L2" - long-name "Chinese traditional (Big5) (Level-2) C940-FEFE" - registries ["big5.eten-0"] - ;; HACK HACK HACK! The `big5' special flag tells the internal code - ;; in Fload_unicode_mapping_table() to take codepoints out of the - ;; Big5 table, convert them to a codepoint in a "fake" chinese-big5-1 - ;; or chinese-big5-2, and store appropriately. Hence, it actually - ;; ignores the name of the charset on which the property is set and - ;; always stores in the "right" place. Correspondingly, we must set - ;; the property on big5-2, not 1, so that both charsets will be - ;; created by the time we initialize the map. - unicode-map ("unicode/unicode-consortium/EASTASIA/OBSOLETE/BIG5.TXT" - nil nil nil big5) - tags (kanji traditional-chinese chinese/language) - )) - ) +;; Big5 claims to be a 94x157 charset, but with gaps in the middle. +;; In particular, the rows are (theoretically) indexed from A1 - FE +;; and the columns from 40 - 7E and A1 - FE. In fact, there are gaps +;; in the rows as well (rows C7 and C8 are missing, as well as rows +;; FA - FE), but that appears to be due to accident -- i.e. they just +;; ran out of chars and/or wanted to make room for expansion. Note +;; also that the gap at C7 and C8 is due to the Level-1/Level-2 +;; division of Big5 (see below). The 94 rows are those between +;; A1 and FE, inclusive. The 157 columns count the sum of the columns +;; in each disjoint set. For us, we need to use the size of the range +;; [40, FE], which is 191. + +;; Old Mule situation; we can only handle up to 96x96 charsets. +;; we split it into two charsets. According to Ken Lunde's CJKV +;; book, Big5 itself is split into "Big Five Level 1" (rows A1-C6) +;; and "Big Five Level 2" (rows C9-F9), with the latter containing +;; less used characters. We split the same way then coerce the +;; result into a 94x94 block. + +;; Since escape-quoted, used for byte-compiled files and auto-saves even on +;; unicode-internal, is ISO 2022-based, it's not practical to switch, as Ben +;; attempted to do, to a single Big5 charset for unicode-internal; doing this +;; would mean byte-complied code from 21.4 would error on 21.5. + +(make-charset + 'chinese-big5-1 + "Frequently used part (A141-C67F) of Big5 (Chinese traditional)" + '(dimension 2 + chars 94 + final ?0 + graphic 0 + short-name "Chinese traditional (Big5), L1" + long-name "Chinese traditional (Big5) (Level-1) A141-C67F" + registries ["big5.eten-0"] + ;; no unicode map, see chinese-big5-2 + tags (kanji traditional-chinese chinese/language))) + +(make-charset + 'chinese-big5-2 + "Less frequently used part (C940-FEFE) of Big5 (Chinese traditional)" + '(dimension 2 + chars 94 + final ?1 + graphic 0 + short-name "Chinese traditional (Big5), L2" + long-name "Chinese traditional (Big5) (Level-2) C940-FEFE" + registries ["big5.eten-0"] + ;; HACK HACK HACK! The `big5' special flag tells the internal code + ;; in Fload_unicode_mapping_table() to take codepoints out of the + ;; Big5 table, convert them to a codepoint in a "fake" chinese-big5-1 + ;; or chinese-big5-2, and store appropriately. Hence, it actually + ;; ignores the name of the charset on which the property is set and + ;; always stores in the "right" place. Correspondingly, we must set + ;; the property on big5-2, not 1, so that both charsets will be + ;; created by the time we initialize the map. + unicode-map ("unicode/unicode-consortium/EASTASIA/OBSOLETE/BIG5.TXT" + nil nil nil big5) + tags (kanji traditional-chinese chinese/language))) ;;;;;;;;;;;;;;;;;;;;; Korean ;;;;;;;;;;;;;;;;;;;; diff -r 19876e056bb7 -r e3925877abce src/ChangeLog --- a/src/ChangeLog Thu Oct 05 00:08:17 2017 +0100 +++ b/src/ChangeLog Fri Oct 06 21:09:22 2017 +0100 ＠＠ -1,3 +1,25 ＠＠ +2017-10-06 Aidan Kehoe <kehoea(a)parhasard.net&gt; + + * lisp.h: + * mule-charset.c: + * mule-charset.c (syms_of_mule_charset): + * mule-charset.c (complex_vars_of_mule_charset): + * mule-charset.c (init_mule_charset): + * mule-coding.c: + * mule-coding.c (byte_big5_two_byte_1_p): + * mule-coding.c (big5_decode): + * mule-coding.c (big5_encode): + * mule-coding.c (decode_big5_char): + * mule-coding.c (big5_char_to_fake_codepoint): + * mule-coding.c (Fencode_big5_char): + * mule-coding.c (init_mule_coding): + * text.c: + * unicode.c (verify_load_unicode_args): + Remove the attempt to implement Big5 as one charset under + unicode-internal; this makes unicode-internal builds incompatible + with byte-code compiled by 21.4 and even source code, + e.g. leim/quail/Punct-B5.el. + 2017-10-04 Aidan Kehoe <kehoea(a)parhasard.net&gt; * text.c (old_mule_ichar_to_unicode): diff -r 19876e056bb7 -r e3925877abce src/lisp.h --- a/src/lisp.h Thu Oct 05 00:08:17 2017 +0100 +++ b/src/lisp.h Fri Oct 06 21:09:22 2017 +0100 ＠＠ -5824,12 +5824,10 ＠＠ Qchinese_cns11643_1, Qchinese_cns11643_2, #ifdef UNICODE_INTERNAL - Qchinese_big5, Qjapanese_shift_jis, -#else /* not UNICODE_INTERNAL */ +#endif /* UNICODE_INTERNAL */ Qchinese_big5_1, Qchinese_big5_2, -#endif /* UNICODE_INTERNAL */ Qchinese_sisheng, Qcomposite; diff -r 19876e056bb7 -r e3925877abce src/mule-charset.c --- a/src/mule-charset.c Thu Oct 05 00:08:17 2017 +0100 +++ b/src/mule-charset.c Fri Oct 06 21:09:22 2017 +0100 ＠＠ -60,12 +60,10 ＠＠ Lisp_Object Vcharset_japanese_jisx0212; Lisp_Object Vcharset_chinese_cns11643_1; Lisp_Object Vcharset_chinese_cns11643_2; -#ifdef UNICODE_INTERNAL -Lisp_Object Vcharset_chinese_big5; -Lisp_Object Vcharset_japanese_shift_jis; -#else Lisp_Object Vcharset_chinese_big5_1; Lisp_Object Vcharset_chinese_big5_2; +#ifdef UNICODE_INTERNAL +Lisp_Object Vcharset_japanese_shift_jis; #endif /* UNICODE_INTERNAL */ Lisp_Object Vcharset_composite; ＠＠ -197,12 +195,10 ＠＠ Qjapanese_jisx0212, Qchinese_cns11643_1, Qchinese_cns11643_2, -#ifdef UNICODE_INTERNAL - Qchinese_big5, - Qjapanese_shift_jis, -#else /* not UNICODE_INTERNAL */ Qchinese_big5_1, Qchinese_big5_2, +#ifdef UNICODE_INTERNAL + Qjapanese_shift_jis, #endif /* UNICODE_INTERNAL */ Qchinese_sisheng, Qcomposite; ＠＠ -1838,12 +1834,10 ＠＠ DEFSYMBOL (Qjapanese_jisx0212); DEFSYMBOL (Qchinese_cns11643_1); DEFSYMBOL (Qchinese_cns11643_2); -#ifdef UNICODE_INTERNAL - DEFSYMBOL (Qchinese_big5); - DEFSYMBOL (Qjapanese_shift_jis); -#else /* not UNICODE_INTERNAL */ DEFSYMBOL (Qchinese_big5_1); DEFSYMBOL (Qchinese_big5_2); +#ifdef UNICODE_INTERNAL + DEFSYMBOL (Qjapanese_shift_jis); #endif /* UNICODE_INTERNAL */ DEFSYMBOL (Qcomposite); ＠＠ -2021,17 +2015,14 ＠＠ Vcharset_chinese_cns11643_1 = Qnil; staticpro (&Vcharset_chinese_cns11643_2); Vcharset_chinese_cns11643_2 = Qnil; -#ifdef UNICODE_INTERNAL - /* We can support Shift-JIS and Big5 directly.*/ - staticpro (&Vcharset_japanese_shift_jis); - Vcharset_japanese_shift_jis = Qnil; - staticpro (&Vcharset_chinese_big5); - Vcharset_chinese_big5 = Qnil; -#else staticpro (&Vcharset_chinese_big5_1); Vcharset_chinese_big5_1 = Qnil; staticpro (&Vcharset_chinese_big5_2); Vcharset_chinese_big5_2 = Qnil; +#ifdef UNICODE_INTERNAL + /* We can support Shift-JIS directly.*/ + staticpro (&Vcharset_japanese_shift_jis); + Vcharset_japanese_shift_jis = Qnil; #endif /* UNICODE_INTERNAL */ #ifdef ENABLE_COMPOSITE_CHARS ＠＠ -2144,12 +2135,10 ＠＠ Vcharset_japanese_jisx0212 = Fget_charset (Qjapanese_jisx0212); Vcharset_chinese_cns11643_1 = Fget_charset (Qchinese_cns11643_1); Vcharset_chinese_cns11643_2 = Fget_charset (Qchinese_cns11643_2); -#ifdef UNICODE_INTERNAL - /* We can support Shift-JIS and Big5 directly.*/ - Vcharset_japanese_shift_jis = Fget_charset (Qjapanese_shift_jis); - Vcharset_chinese_big5 = Fget_charset (Qchinese_big5); -#else Vcharset_chinese_big5_1 = Fget_charset (Qchinese_big5_1); Vcharset_chinese_big5_2 = Fget_charset (Qchinese_big5_2); +#ifdef UNICODE_INTERNAL + /* We can support Shift-JIS directly.*/ + Vcharset_japanese_shift_jis = Fget_charset (Qjapanese_shift_jis); #endif /* UNICODE_INTERNAL */ } diff -r 19876e056bb7 -r e3925877abce src/mule-coding.c --- a/src/mule-coding.c Thu Oct 05 00:08:17 2017 +0100 +++ b/src/mule-coding.c Fri Oct 06 21:09:22 2017 +0100 ＠＠ -898,8 +898,6 ＠＠ /* BIG5 (used for Mandarin in Taiwan). */ DEFINE_CODING_SYSTEM_TYPE_WITH_DATA (big5); -#ifndef UNICODE_INTERNAL - /* BIG5 is a coding system encoding two character sets: ASCII and Big5. An ASCII character is encoded as is. Big5 is a two-byte character set and is encoded in two-byte. ＠＠ -991,8 +989,6 ＠＠ b2 += b2 < 0x3F ? 0x40 : 0x62; \ } while (0) -#endif /* not UNICODE_INTERNAL */ - inline static int byte_big5_two_byte_1_p (int c) { ＠＠ -1024,17 +1020,11 ＠＠ /* Previous character was first byte of Big5 char. */ if (byte_big5_two_byte_2_p (c)) { -#ifdef UNICODE_INTERNAL - non_ascii_charset_codepoint_to_dynarr - (Vcharset_chinese_big5, data->ch, c, dst, - CONVERR_USE_PRIVATE); -#else /* not UNICODE_INTERNAL */ Lisp_Object charset; int b1, b2; DECODE_BIG5 (data->ch, c, charset, b1, b2); non_ascii_charset_codepoint_to_dynarr (charset, b1, b2, dst, CONVERR_USE_PRIVATE); -#endif /* UNICODE_INTERNAL */ } else { ＠＠ -1084,13 +1074,6 ＠＠ } ichar_to_charset_codepoint (ich, Vbig5_precedence, &charset, &c1, &c2, CONVERR_FAIL); -#ifdef UNICODE_INTERNAL - if (EQ (charset, Vcharset_chinese_big5)) - { - Dynarr_add (dst, c1); - Dynarr_add (dst, c2); - } -#else /* not UNICODE_INTERNAL */ if (EQ (charset, Vcharset_chinese_big5_1) || EQ (charset, Vcharset_chinese_big5_2)) { ＠＠ -1099,7 +1082,6 ＠＠ Dynarr_add (dst, b1); Dynarr_add (dst, b2); } -#endif /* UNICODE_INTERNAL */ else { handle_standard_encoding_error (str, src, dst); ＠＠ -1131,10 +1113,6 ＠＠ static Ichar decode_big5_char (int b1, int b2) { -#ifdef UNICODE_INTERNAL - return charset_codepoint_to_ichar (Vcharset_chinese_big5, b1, b2, - CONVERR_FAIL); -#else /* not UNICODE_INTERNAL */ if (byte_big5_two_byte_1_p (b1) && byte_big5_two_byte_2_p (b2)) { ＠＠ -1146,11 +1124,8 ＠＠ } else return -1; -#endif /* UNICODE_INTERNAL */ } -#ifndef UNICODE_INTERNAL - void big5_char_to_fake_codepoint (int b1, int b2, Lisp_Object *charset, int *c1, int *c2) ＠＠ -1168,8 +1143,6 ＠＠ } } -#endif /* not UNICODE_INTERNAL */ - DEFUN ("decode-big5-char", Fdecode_big5_char, 1, 1, 0, /* Convert Big Five character codes in CODE into a character. CODE is a cons of two integers specifying the codepoints in Big Five. ＠＠ -1208,12 +1181,6 ＠＠ CHECK_CHAR_COERCE_INT (character); ichar_to_charset_codepoint (XCHAR (character), Vbig5_precedence, &charset, &c1, &c2, CONVERR_FAIL); -#ifdef UNICODE_INTERNAL - if (EQ (charset, Vcharset_chinese_big5)) - { - return Fcons (make_fixnum (c1), make_fixnum (c2)); - } -#else /* not UNICODE_INTERNAL */ if (EQ (charset, Vcharset_chinese_big5_1) || EQ (charset, Vcharset_chinese_big5_2)) { ＠＠ -1221,7 +1188,6 ＠＠ ENCODE_BIG5 (charset, c1, c2, b1, b2); return Fcons (make_fixnum (b1), make_fixnum (b2)); } -#endif /* UNICODE_INTERNAL */ else return Qnil; } ＠＠ -4187,14 +4153,7 ＠＠ simple_convert_predence_list_to_array (list3 (Vcharset_japanese_jisx0208, Vcharset_japanese_jisx0208_1978, Vcharset_katakana_jisx0201)); - -#ifdef UNICODE_INTERNAL - Vbig5_precedence = - simple_convert_predence_list_to_array - (list1 (Vcharset_chinese_big5)); -#else /* not UNICODE_INTERNAL */ Vbig5_precedence = simple_convert_predence_list_to_array (list2 (Vcharset_chinese_big5_1, Vcharset_chinese_big5_2)); -#endif /* UNICODE_INTERNAL */ } diff -r 19876e056bb7 -r e3925877abce src/text.c --- a/src/text.c Thu Oct 05 00:08:17 2017 +0100 +++ b/src/text.c Fri Oct 06 21:09:22 2017 +0100 ＠＠ -5521,14 +5521,11 ＠＠ All codes above are two decimal numbers except for Big Five and ANSI Z39.64, which we don't support. We add 32 to each of the decimal numbers. -Note that in an old-Mule world (see below), Big Five is split in a rather -hackish fashion into two charsets, `chinese-big5-1' and `chinese-big5-2', -due to its excessive size (94x157), with the first codepoint in the range -0xA1 to 0xFE and the second in the range 0x40 to 0x7E or 0xA1 to 0xFE. -`decode-big5-char' is used to generate the char from its codes, and -`encode-big5-char' extracts the codes. This hack doesn't exist with -Unicode-internal, and hence the expression (make-char 'chinese-big5 164 116) -could be used. +Big Five is split in a rather hackish fashion into two charsets, +`chinese-big5-1' and `chinese-big5-2', due to its excessive size (94x157), +with the first codepoint in the range 0xA1 to 0xFE and the second in the range +0x40 to 0x7E or 0xA1 to 0xFE. `decode-big5-char' is used to generate the char +from its codes, and `encode-big5-char' extracts the codes. Note that there are three different internal formats for characters: diff -r 19876e056bb7 -r e3925877abce src/unicode.c --- a/src/unicode.c Thu Oct 05 00:08:17 2017 +0100 +++ b/src/unicode.c Fri Oct 06 21:09:22 2017 +0100 ＠＠ -2248,10 +2248,8 ＠＠ { if (EQ (elt, Qignore_first_column)) *flags_out |= LOAD_UNICODE_IGNORE_FIRST_COLUMN; -#ifndef UNICODE_INTERNAL else if (EQ (elt, Qbig5)) *flags_out |= LOAD_UNICODE_BIG5; -#endif /* not UNICODE_INTERNAL */ else invalid_constant ("Unrecognized `load-unicode-mapping-table' flag", elt); ＠＠ -2324,9 +2322,7 ＠＠ int flgs; int stage; int to_unicode_min_val[256], to_unicode_max_val[256]; -#ifndef UNICODE_INTERNAL int big5_other_unicode_min_val[256], big5_other_unicode_max_val[256]; -#endif int i; /* This may be called to autoload the Unicode tables, from a function ＠＠ -2338,7 +2334,6 ＠＠ verify_load_unicode_args (filename, start, end, offset, flags, &st, &en, &of, &flgs); -#ifndef UNICODE_INTERNAL if (flgs & LOAD_UNICODE_BIG5) { /* At this point the charsets haven't been initialized ＠＠ -2353,7 +2348,6 ＠＠ big5_other_unicode_max_val[i] = 0; } } -#endif /* not UNICODE_INTERNAL */ for (i = 0; i < 256; i++) { ＠＠ -2482,7 +2476,6 ＠＠ cp1high = cp1 >> 8; cp1low = cp1 & 255; -#ifndef UNICODE_INTERNAL if (flgs & LOAD_UNICODE_BIG5) { big5_char_to_fake_codepoint (cp1high, cp1low, ＠＠ -2509,7 +2502,6 ＠＠ set_unicode_conversion (cp2, charset, c1, c2); } else -#endif /* not UNICODE_INTERNAL */ { int l1, l2, h1, h2; c1 = cp1high, c2 = cp1low; ＠＠ -2520,9 +2512,7 ＠＠ goto out_of_range; } -#ifndef UNICODE_INTERNAL do_it: -#endif if (stage == 0) { if (c2 < to_unicode_min_val[c1]) -- ‘As I sat looking up at the Guinness ad, I could never figure out / How your man stayed up on the surfboard after forty pints of stout’ (C. Moore)