LC_MESSAGES problem ?

[Won-kyu Park]

On Mon, Nov 20, 2000 at 08:55:54AM -0800, Ulrich Drepper wrote:
> Won-kyu Park <wkpark@chem.skku.ac.kr> writes:
> 
> > But, Why the default charset is ASCII ? translation messages have it's
> > own charset info, like 'charset=EUC-KR'.
> > 
> > IMHO, the default guessed charset by intl/loadmsgcat.c have to be fixed by
> > proper manner.
> 
> No.  The terminal used (represented by the LC_CTYPE locale) and the
> message catalog selected need not use the same character set.
>

There are many programs do not use setlocale(LC_CTYPE,"").
some programs are LC_CTYPE sensitive (like isctype() etc.) and then
set the LC_CTYPE as 'C'. but... how can I get a translation messages
under glibc ?
(with OUTPUT_CHARSET ?)
-----------
...
main() {
...
   setlocale(LC_CTYPE,"C");
   setlocale(LC_MESSAGES,"");
...
-----------

$ LC_MESSAGES=ko_KR ./a.out
???? ?????

Solaris works fine with setlocale(LC_MESSAGES,"") only.
(without setlocale(LC_CTYPE,""))

There are many programs using setlocale(LC_MESSAES,"") only.
(the gcc also)

With this 'LC_MESSAGES' problem, Linux users confused with '????? ???'
messages at the East Asia.