bug in srfi-13

bug in srfi-13

[Han-Wen Nienhuys]

test.scm:

	(use-modules (srfi srfi-13))
	(string-map
	 (lambda (chr) (display (list " " (char->integer chr))) chr)
	"ö")

=>

	byrd:~/usr/src/lilypond$ guile test.scm
	(  16777155)(  16777142)

after bugfix:

	byrd:~/usr/src/lilypond$ guile test.scm
	(  195)(  182)

there is still something broken in this function. For reasons beyond
my comprehension, the SCM_VALIDATE_SUBSTRING_SPEC_COPY macro has the
side effect of assigning to cstart and cend, and doing so wrongly.

May I suggest that future SCM_VALIDATE_*_  macros be renamed, for
example from

	SCM_VALIDATE_COPY

to

	SCM_VALIDATED_COPY

The latter implies that the macro does have a side effect.

Also, the naming of scm_{up,down}case functions was wrong; it should
be scm_c_{up,down}case.  I have changed this. I leave to Marius to
decide whether a backward compatibility must be maintained.

I am still puzzled why <ctype.h> isn't used.



-- 
 Han-Wen Nienhuys   |   hanwen@xs4all.nl   |   http://www.xs4all.nl/~hanwen 



_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

bug in srfi-13

[Han-Wen Nienhuys]

hanwen@xs4all.nl writes:
> 
> after bugfix:
> 
> 	byrd:~/usr/src/lilypond$ guile test.scm
> 	(  195)(  182)
> 
> there is still something broken in this function. For reasons beyond
> my comprehension, the SCM_VALIDATE_SUBSTRING_SPEC_COPY macro has the
> side effect of assigning to cstart and cend, and doing so wrongly.
> 

I retract this. Emacs expands the o with dieresis to two bytes
characters;  my bad. 

-- 

 Han-Wen Nienhuys   |   hanwen@xs4all.nl   |   http://www.xs4all.nl/~hanwen 



_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

Re: bug in srfi-13

[Kevin Ryde]

Han-Wen Nienhuys  <hanwen@xs4all.nl> writes:
>
> after bugfix:
>
> 	byrd:~/usr/src/lilypond$ guile test.scm
> 	(  195)(  182)

That fix might be a candidate for the 1.6 branch too.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

Re: bug in srfi-13

[Kevin Ryde]

Han-Wen Nienhuys  <hanwen@xs4all.nl> writes:
>
> I am still puzzled why <ctype.h> isn't used.

It looks a bit suss that char-upcase and char-downcase should do their
own thing, but char-upper-case? and char-lower-case? use ctype.h.

There's probably some good scope for inconsistency there in an 8-bit
locale.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

Re: bug in srfi-13

[Kevin Ryde]

[-- Attachment #1: Type: text/plain, Size: 595 bytes --]

I'd be thinking of something like this for char-upcase etc,

        * chars.c (scm_char_upcase, scm_char_downcase, scm_c_upcase,
        scm_c_downcase): Use ctype.h toupper and tolower.  This will be useful
        in 8-bit locales, and ensures consistency with char-upper-case? and
        char-lower-case? which already use ctype.h.
        (scm_c_upcase_table, scm_c_downcase_table, scm_lowers, scm_uppers):
        Remove.
        * chars.c, chars.h, init.c (scm_tables_prehistory): Remove.

This would be for 1.6 too, the dodigness between using and not using
ctype.h is the same there.


[-- Attachment #2: chars.c.toupper.diff --]
[-- Type: text/plain, Size: 2120 bytes --]

--- chars.c.~1.35.~	2004-04-15 10:58:00.000000000 +1000
+++ chars.c	2004-04-18 10:18:19.000000000 +1000
@@ -1,4 +1,4 @@
-/*	Copyright (C) 1995,1996,1998, 2000, 2001 Free Software Foundation, Inc.
+/*	Copyright (C) 1995,1996,1998, 2000, 2001, 2004 Free Software Foundation, Inc.
  * 
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
@@ -19,6 +19,7 @@
 \f
 
 #include <ctype.h>
+#include <limits.h>
 #include "libguile/_scm.h"
 #include "libguile/validate.h"
 
@@ -260,7 +261,7 @@
 #define FUNC_NAME s_scm_char_upcase
 {
   SCM_VALIDATE_CHAR (1, chr);
-  return SCM_MAKE_CHAR(scm_c_upcase (SCM_CHAR(chr)));
+  return SCM_MAKE_CHAR (toupper (SCM_CHAR (chr)));
 }
 #undef FUNC_NAME
 
@@ -271,7 +272,7 @@
 #define FUNC_NAME s_scm_char_downcase
 {
   SCM_VALIDATE_CHAR (1, chr);
-  return SCM_MAKE_CHAR(scm_c_downcase (SCM_CHAR(chr)));
+  return SCM_MAKE_CHAR (tolower (SCM_CHAR(chr)));
 }
 #undef FUNC_NAME
 
@@ -279,25 +280,6 @@
 
 
 
-static unsigned char scm_c_upcase_table[SCM_CHAR_CODE_LIMIT];
-static unsigned char scm_c_downcase_table[SCM_CHAR_CODE_LIMIT];
-static const unsigned char scm_lowers[] = "abcdefghijklmnopqrstuvwxyz";
-static const unsigned char scm_uppers[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ";
-
-
-void 
-scm_tables_prehistory ()
-{
-  int i;
-  for (i = 0; i < SCM_CHAR_CODE_LIMIT; i++)
-    scm_c_upcase_table[i] = scm_c_downcase_table[i] = i;
-  for (i = 0; i < (int) (sizeof scm_lowers / sizeof (scm_lowers[0])); i++)
-    {
-      scm_c_upcase_table[scm_lowers[i]] = scm_uppers[i];
-      scm_c_downcase_table[scm_uppers[i]] = scm_lowers[i];
-    }
-}
-
 /*
 TODO: change name  to scm_i_.. ? --hwn
 */
@@ -306,8 +288,8 @@
 int
 scm_c_upcase (unsigned int c)
 {
-  if (c < sizeof (scm_c_upcase_table))
-    return scm_c_upcase_table[c];
+  if (c <= UCHAR_MAX)
+    return toupper (c);
   else
     return c;
 }
@@ -316,8 +298,8 @@
 int
 scm_c_downcase (unsigned int c)
 {
-  if (c < sizeof (scm_c_downcase_table))
-    return scm_c_downcase_table[c];
+  if (c <= UCHAR_MAX)
+    return tolower (c);
   else
     return c;
 }

[-- Attachment #3: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

Re: bug in srfi-13

[Kevin Ryde]

I wrote:
>
>         * chars.c (scm_char_upcase, ...

I applied this to the head, but the boss of 1.6 can say about there.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

Re: bug in srfi-13

[Marius Vollmer]

Kevin Ryde <user42@zip.com.au> writes:

> I applied this to the head, but the boss of 1.6 can say about there.

Yes, please apply this to 1.6 as well.  Thanks!

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

Re: bug in srfi-13

[Marius Vollmer]

Kevin Ryde <user42@zip.com.au> writes:

> This would be for 1.6 too, the dodigness between using and not using
> ctype.h is the same there.

Yes, well done.  Thanks!

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel