Nothing too complex, and even the fragile test suite passes. Here are some profiling numbers, where charsets-in-region is the implementation in C and charsets-in-region0 is the implementation in Lisp: 10,000 calls in *scratch*, vanilla startup Function Name Ticks/Total %Usage Calls GC-Usage/ Total ===============================/===== ====== ===== ========/======= charsets-in-region 74/ 88 77.083 10000 99576/ 120000 (profile overhead) 18/ 18 18.750 20424/ 0 charsets-in-region-*-10000 4/ 96 4.167 1 40/ 120040 #<compiled-function (expr &optional arg) "...(36)" [was-profiling arg expr profiling-active-p ((byte-code ¬~Á ~Á~ [was-profiling stop-profiling] 1)) start-profiling eval profile-results clear-profiling-info ((stop-profiling))] 2 (/usr/local/lib/xemacs/xemacs-packages/lisp/xemacs-devel/profile.elc . 9000) (list (read (read-string Expression to profile: )) current-prefix-arg)> 0/ 96 0.000 0/ 120028 Function Name Ticks/Total %Usage Calls GC-Usage/ Total ================================/===== ====== ===== ========/======= charsets-in-region0 1999/ 2304 89.281 10000 117000/ 120064 (profile overhead) 128/ 128 5.717 3092/ 0 char-charset 110/ 92 4.913 3950000 charsets-in-region0-*-10000 2/ 2306 0.089 1 0/ 120104 #<compiled-function (expr &optional arg) "...(36)" [was-profiling arg expr profiling-active-p ((byte-code ¬~Á ~Á~ [was-profiling stop-profiling] 1)) start-profiling eval profile-results clear-profiling-info ((stop-profiling))] 2 (/usr/local/lib/xemacs/xemacs-packages/lisp/xemacs-devel/profile.elc . 9000) (list (read (read-string Expression to profile: )) current-prefix-arg)> 10,000 calls on HELLO, vanilla startup Function Name Ticks/Total %Usage Calls GC-Usage/ Total ====================================/===== ====== ===== ========/======= charsets-in-region 2620/ 3351 80.690 10000 2040000/2040000 (in char-byte conversion) 576/ 514 17.739 11570000 (in garbage collection) 47/ 47 1.447 1 charsets-in-region-*-10000 4/ 3403 0.123 1 24/2040024 #<compiled-function (expr &optional arg) "...(36)" [was-profiling arg expr profiling-active-p ((byte-code ¬ƒÁ ˆÁ‡ [was-profiling stop-profiling] 1)) start-profiling eval profile-results clear-profiling-info ((stop-profiling))] 2 (/usr/local/lib/xemacs/xemacs-packages/lisp/xemacs-devel/profile.elc . 9000) (list (read (read-string Expression to profile: )) current-prefix-arg)> (a call with charsets-in-region0 crashed in the profiler) 5,000 calls in HELLO, vanilla startup Function Name Ticks/Total %Usage Calls GC-Usage/ Total ====================================/===== ====== ===== ========/======= charsets-in-region 2567/ 3332 79.969 10000 2040000/2040000 (in char-byte conversion) 568/ 479 17.695 11570000 (in garbage collection) 43/ 43 1.340 1 (profile overhead) 28/ 28 0.872 charsets-in-region-*-5000 4/ 3379 0.125 1 64/2040064 Function Name Ticks/Total %Usage Calls GC-Usage/ Total ====================================/===== ====== ===== ========/======= charsets-in-region0 11693/13187 91.295 10000 2041944/2041944 (in char-byte conversion) 735/ 597 5.739 23450000 char-charset 326/ 246 2.545 11740000 (in garbage collection) 46/ 46 0.359 1 charsets-in-region0-*-5000 8/13195 0.062 1 40/2041984 lisp/ChangeLog addition: 2006-11-14 Aidan Kehoe <kehoea(a)parhasard.net&gt; * mule/mule-charset.el: * mule/mule-charset.el (charsets-in-string): Implement it in terms of charsets-in-string. * mule/mule-charset.el (charsets-in-region): Removed. It's now in C. src/ChangeLog addition: 2006-11-14 Aidan Kehoe <kehoea(a)parhasard.net&gt; * mule-charset.c: * mule-charset.c (Fcharsets_in_region): Added a charsets-in-region implementation in C. XEmacs Trunk source patch: Diff command: cvs -q diff -u Files affected: src/text.c =================================================================== RCS src/mule-charset.c =================================================================== RCS lisp/mule/mule-charset.el =================================================================== RCS Index: lisp/mule/mule-charset.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/mule/mule-charset.el,v retrieving revision 1.19 diff -u -r1.19 mule-charset.el --- lisp/mule/mule-charset.el 2006/11/05 22:31:38 1.19 +++ lisp/mule/mule-charset.el 2006/11/14 13:49:48 ＠＠ -38,42 +38,16 ＠＠  ;;;; Classifying text according to charsets -;; the old version was broken in a couple of ways -;; this is one of several versions, I tried a hash as well as the -;; `prev-charset' cache used in the old version, but this was definitely -;; faster than the hash version and marginally faster than the prev-charset -;; version -;; #### this really needs to be moved into C -(defun charsets-in-region (start end &optional buffer) - "Return a list of the charsets in the region between START and END. -BUFFER defaults to the current buffer if omitted." - (let (list) - (save-excursion - (if buffer - (set-buffer buffer)) - (save-restriction - (narrow-to-region start end) - (goto-char (point-min)) - (while (not (eobp)) - ;; the first test will usually succeed on testing the - ;; car of the list; don't waste time let-binding. - (or (memq (char-charset (char-after (point))) list) - (setq list (cons (char-charset (char-after (point))) list))) - (forward-char)))) - list)) - (defun charsets-in-string (string) "Return a list of the charsets in STRING." - (let (list) - (mapc (lambda (ch) - ;; the first test will usually succeed on testing the - ;; car of the list; don't waste time let-binding. - (or (memq (char-charset ch) list) - (setq list (cons (char-charset ch) list)))) - string) - list)) + (let (res) + (with-string-as-buffer-contents string + ;; charsets-in-region now in C. + (setq res (charsets-in-region (point-min) (point-max)))) + res)) (defalias 'find-charset-string 'charsets-in-string) + (defalias 'find-charset-region 'charsets-in-region)  Index: src/mule-charset.c =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/src/mule-charset.c,v retrieving revision 1.51 diff -u -r1.51 mule-charset.c --- src/mule-charset.c 2006/11/12 13:40:08 1.51 +++ src/mule-charset.c 2006/11/14 13:49:48 ＠＠ -937,6 +937,39 ＠＠ return Qnil; } +DEFUN ("charsets-in-region", Fcharsets_in_region, 2, 3, 0, /* +Return a list of the charsets in the region between START and END. +BUFFER defaults to the current buffer if omitted. +*/ + (start, end, buffer)) +{ + /* This function can GC */ + struct buffer *buf = decode_buffer (buffer, 1); + Charbpos pos, stop; /* Limits of the region. */ + Lisp_Object res = Qnil; + int charsets[NUM_LEADING_BYTES]; + Ibyte lb; + struct gcpro gcpro1; + + memset(charsets, 0, sizeof(charsets)); + get_buffer_range_char (buf, start, end, &pos, &stop, 0); + + GCPRO1 (res); + while (pos < stop) + { + lb = ichar_leading_byte(BUF_FETCH_CHAR (buf, pos)); + if (0 == charsets[lb - MIN_LEADING_BYTE]) + { + charsets[lb - MIN_LEADING_BYTE] = 1; + res = Fcons (XCHARSET_NAME(charset_by_leading_byte(lb)), res); + } + ++pos; + } + UNGCPRO; + + return res; +} +  /************************************************************************/ /* memory usage */ ＠＠ -1029,6 +1062,7 ＠＠ DEFSUBR (Fcharset_id); DEFSUBR (Fset_charset_ccl_program); DEFSUBR (Fset_charset_registries); + DEFSUBR (Fcharsets_in_region); #ifdef MEMORY_USAGE_STATS DEFSUBR (Fcharset_memory_usage); -- Santa Maradona, priez pour moi! _______________________________________________ XEmacs-Patches mailing list XEmacs-Patches(a)xemacs.org http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-patches