Re: [bug-gnu-libiconv] [PATCH] Support nl_langinfo (CODESET) correctly on OS/2

[Bruno Haible]

Hi KO,

> diff --git a/libcharset/lib/localcharset.c b/libcharset/lib/localcharset.c
> index da3ac45..40923fc 100644
> --- a/libcharset/lib/localcharset.c
> +++ b/libcharset/lib/localcharset.c
> @@ -378,26 +378,41 @@ static const struct table_entry alias_table[] =
>         by Alex Taylor:
>         <http://altsan.org/os2/toolkits/uls/index.html#codepages>.
>         See also "IBM Globalization - Code page identifiers":
> -       <https://www-01.ibm.com/software/globalization/cp/cp_cpgid.html>.  */
> -    { "CP1089", "ISO-8859-6" },
> -    { "CP1208", "UTF-8" },
> -    { "CP1381", "GB2312" },
> -    { "CP1386", "GBK" },
> -    { "CP3372", "EUC-JP" },
> -    { "CP813",  "ISO-8859-7" },
> -    { "CP819",  "ISO-8859-1" },
> -    { "CP878",  "KOI8-R" },
> -    { "CP912",  "ISO-8859-2" },
> -    { "CP913",  "ISO-8859-3" },
> -    { "CP914",  "ISO-8859-4" },
> -    { "CP915",  "ISO-8859-5" },
> -    { "CP916",  "ISO-8859-8" },
> -    { "CP920",  "ISO-8859-9" },
> -    { "CP921",  "ISO-8859-13" },
> -    { "CP923",  "ISO-8859-15" },
> -    { "CP954",  "EUC-JP" },
> -    { "CP964",  "EUC-TW" },
> -    { "CP970",  "EUC-KR" }
> +       <https://www-01.ibm.com/software/globalization/cp/cp_cpgid.html>.

This URL is dead. You can remove it, or replace it with another suitable one.

> +       See also "__convcp() of kLIBC":
> +       
> <http://trac.netlabs.org/libc/browser/branches/libc-0.6/src/emx/src/lib/locale/__convcp.c>,
> +       or:
> +       
> <https://github.com/bitwiseworks/libc/blob/master/src/emx/src/lib/locale/__convcp.c>.
>   */

The first of these two URLs is broken. You can therefore remove it.

> +    { "CP1089",         "ISO-8859-6" },

OK

> +    { "CP1200",         "UCS-2" },

UCS-2 cannot be used as a locale encoding, since it is not ASCII compatible. 
This
cannot work.

> +    { "CP1208",         "UTF-8" },
> +    { "CP1381",         "GB2312" },
> +    { "CP1383",         "EUC-CN" },
> +    { "CP1386",         "GBK" },
> +    { "CP3372",         "EUC-JP" },
> +    { "CP813",          "ISO-8859-7" },
> +    { "CP819",          "ISO-8859-1" },
> +    { "CP878",          "KOI8-R" },
> +    { "CP912",          "ISO-8859-2" },
> +    { "CP913",          "ISO-8859-3" },
> +    { "CP914",          "ISO-8859-4" },
> +    { "CP915",          "ISO-8859-5" },
> +    { "CP916",          "ISO-8859-8" },
> +    { "CP920",          "ISO-8859-9" },
> +    { "CP921",          "ISO-8859-13" },
> +    { "CP923",          "ISO-8859-15" },
> +    { "CP954",          "EUC-JP" },
> +    { "CP964",          "EUC-TW" },
> +    { "CP970",          "EUC-KR" },
> +    { "ISO8859-1",      "ISO-8859-1" },
> +    { "ISO8859-2",      "ISO-8859-2" },
> +    { "ISO8859-3",      "ISO-8859-3" },
> +    { "ISO8859-4",      "ISO-8859-4" },
> +    { "ISO8859-5",      "ISO-8859-5" },
> +    { "ISO8859-6",      "ISO-8859-6" },
> +    { "ISO8859-7",      "ISO-8859-7" },
> +    { "ISO8859-8",      "ISO-8859-8" },
> +    { "ISO8859-9",      "ISO-8859-9" }

OK

> @@ -751,6 +766,24 @@ locale_charset (void)
>      }
>  #  endif
>  
> +#  ifdef OS2
> +  /* On OS/2, nl_langinfo (CODESET) returns IBM-XXX style normally. Convert 
> it
> +     to CPXXX style for mapping later except UCS-2LE and UCS-2BE.  */
> +  if (strcmp (codeset, "IBM-1200@endian=little") == 0)
> +    return "UCS-2LE";
> +  else if (strcmp (codeset, "IBM-1200@endian=big") == 0)
> +    return "UCS-2BE";
> +
> +  if (strncmp (codeset, "IBM-", 4) == 0 && isdigit (codeset[4]))
> +    {
> +      static char buf[2 + 10 + 1];
> +
> +      snprintf (buf, sizeof (buf), "CP%s", codeset + 4);
> +
> +      codeset = buf;
> +    }
> +#  endif
> +

I would prefer if you could get rid of this extra code, and instead just add
entries to the table above.

UCS-2LE and UCS-2BE are also unsuitable for locale encodings.

Bruno