From: Emmanuel Blot (emmanuel.blot@sun.com)
Date: Fri Nov 23 2001 - 08:47:11 CST
>
>
>
>>+#ifdef HAVE_LOCALE_H
>>+ if ( ! setlocale(LC_ALL, set_language) ) {
>>+ sprintf(errmsg, "WARNING: setlocale: \"%s\" %s.", set_language,
>>+ lang[MSG_LANGUAGE_NOT_SUPPORTED]);
>>+ fprintf(stderr, errmsg);
>>+ }
>>+#endif
>>
>
> I get this warning when I run it:
>WARNING: setlocale: "en" Language not supported.
>
Ok, I found the problem:
Several system are relying on a more complex locale definition, that is,
locale needs two parameters:
* language
* country
that is, locale parameter is ll_cc[opt]
where:
ll is the language in ISO language format, lowercase
cc is the country in ISO country format, uppercase
opt (optional parameters) may vary, for example it can be "@euro"
for example, english language can be set as:
en_GB
en_IE
en_US
etc. You can find the complete list available on your system in
/usr/share/i18n/locales
How to deal with such a parameter in hypermail ?
1/ use two parameters: one for the message language, one for the
LC_TIME localization
I don't really like this idea. It's not straightforward for the
end user
2/ use one parameters, in ISO format:
en_GB or fr_FR for example
It could be easily achieve this way:
#ifdef HAVE_LOCALE_H
#use the full ISO locale for system
if ( ! setlocale(LC_TIME /*or LC_ALL */, set_language) ) {
sprintf(errmsg, "WARNING: setlocale: \"%s\" %s.", set_language,
lang[MSG_LANGUAGE_NOT_SUPPORTED]);
fprintf(stderr, errmsg);
}
#endif
# restrict locale to the language parameter, remove country and
extra paramters if any
if ( strlen(set_language) > 2 && set_language[2] == '_' ) {
set_language[2] = '\0';
}
# the old code is not changed
if ((tlang = valid_language(set_language)) == NULL) {
sprintf(errmsg, "\"%s\" %s.", set_language,
lang[MSG_LANGUAGE_NOT_SUPPORTED]);
cmderr(errmsg);
}
It also requires a smal effort on the help message, and an updated
FAQ/documentation, in order to produce something like this:
> -L lang : Specify language to use (de_* en_* es_* fi_* fr_*
is_* se_* )
(note that the se_ parameter has still to be changed to match the ISO
format: sv_*)
I think this new implementation will fix up the problem AND provide
'standard' localization.
Regards,
Emmanuel.