Re: gtk i18n message files

[Pablo Saratxaga <pablo mandrakesoft com>]

Kaixo!

On Tue, Sep 04, 2001 at 11:57:49AM -0700, keyur shroff wrote:

> hi_IN locale in C library. Now I can see Hindi text
> when I set
>    export LANG=hi_IN.UTF-8

You can?
It was my understanding that support of devanagari script was possible
only with gtk 2.

> Now comes to my problem. I have created message files
> (.po files) for varius gtk applications where message
> strings are stored using UTF-8 encoding. From this .po
> files I generated a number of .mo files and put it in
> the LC_MESSAGES directory of current locale path. Now
> when gettext interface reads these messages, it again
> converts each byte of UTF-8 Hindi text into multibyte
> sequence.

That means your po files are wrong. Check that the header decalres
charset=utf-8.

> method, I get UTF-8 sequence for each of the bytes E0,
> A4, and 95,
>  i.e., C3 A0  C2 A4  A2 95
>        =====  =====  =====
>          E0     A4     95

Probably youare telling in the po file it is in iso-8859-1 encoding;
when you request UTF-8 trough your locale setting, then gettext does
what you ask: convert iso-8859-1 to utf-8.

Fix the po files.

> As a result Hindi text is not displayed at all. Can
> someone tell me what may be the problem? Is this the
> problem with glibc? 

No, it is a problem with the po file being wrong. 

> PS : Please note that I am using proper fonts in
> fontset and I can see 'day of the week' and 'month
> names' in Hindi in various gtk applications. 

the output of strftime and similar is done by the libc itself, not
by any po file.


-- 
Ki ça vos våye bén,
Pablo Saratxaga

http://www.srtxg.easynet.be/		PGP Key available, key ID: 0x8F0E4975