Bug#183515: locales: busted postinst script?

[GOTO Masanori <gotom@debian.or.jp>]

At Sat, 08 Mar 2003 15:25:08 +0900,
Stephen J. Turnbull <stephen@xemacs.org> wrote:
> 
> >>>>> "GOTO" == GOTO Masanori <gotom@debian.or.jp> writes:
> 
>     GOTO> BTW, did you install "localeconf"?  localeconf sometimes has
>     GOTO> caused problems with locales.  They don't consider locales
>     GOTO> well and not communicate with us.
> 
> I don't think so:
> 
> bash-2.05b$ dpkg -l localeconf
> No packages found matching localeconf.
> bash-2.05b$ which localeconf
> bash-2.05b$ 

That's OK, thanks.

>     GOTO> /etc/locale.gen is a simple format file, so it might be
>     GOTO> difficult to detect the wrong format data.  However, at
>     GOTO> least it does not generate an invalid locale data with wrong
>     GOTO> locale name...
> 
> Well, since localedef (I think it was) barfed a message it probably
> also gives an error return.  The return code should be checked, that's
> easy.

So if we check the return code of localedef, and it says error, then
stop to generate locale data?  I wouldn't like to do so, because
sometimes localedata itself is broken.  For example, LC_MONETARY
(monetary category) of sr_YU (Yugoslavia) locale is broken for a long
time because the abbreviated currency symbol of Yugoslav (YUM or YUN)
is not decided well (upstream makes it pending state).  It's difficult
to avoid stopping to generate locale data because of such a
(sometimes) political reason.

If user's first entry of /etc/locale.gen is sr_YU, and the second is
en_US, so he/she can't use en_US locale because locale-gen stops in
the way.

Regards,
-- gotom