Re: non-utf8 po files breaking the build

[Kjartan Maraas <kmaraas online no>]

søn, 2002-11-24 kl. 21:12 skrev Keld Jørn Simonsen:
> On Sun, Nov 24, 2002 at 05:14:25PM +0100, Christian Rose wrote:
> > sön 2002-11-24 klockan 17.08 skrev Yanko Kaneti:
> > > > Maybe this could be modified a bit, so that po files that are already
> > > > valid UTF-8 won't be unnecessarily touched?
> > > 
> > > ok here is the final version.
> > > 
> > > for F in `grep -l 'charset=[^U]' *.po` ; do msgconv -t UTF-8 $F -o $F.utf && mv $F.utf $F; done
> > > 
> > > Files differing only by their timestamp dont get committed anyway.
> > > msgconv takes care of changing charset=utf-8 to charset=UTF-8.
> > 
> > I think this would be mostly ok, but perhaps inverse matching on a full
> > "charset=UTF-8" would be more failsafe than just inverse matching on an
> > initial U in the charset.
> 
> I don't think you should do it, it would create some burdens for 
> translators that are unnecesary.
> 

There are lots of things that are unneccesary in the world :-)

I think the burden that is placed on the maintainers and bug managers
outweighs the burden placed on the translators in this case. If we could
manage to get it set up so that the files are converted automatically on
checkin there's absolutely no burden placed on the translators at all.

Not having all po-files in UTF-8 is causing real problems every week.
Even it is because the tools are bad at handling every corner case when
dealing with all kinds of different encodings I think it's easier to
just say "Translations shall be in UTF-8" than to fix all bugs in the
tools. I switched to UTF-8 myself a long time ago and I can't really say
that it has caused any problems for me...

Cheers
Kjartan