Re: UTF-8

[Christian Rose <menthos menthos com>]

mån 2002-07-08 klockan 23.18 skrev Keld Jørn Simonsen:
> > > > > Why do they need to be in utf8?
> > > >
> > > > I think we'd like gnomemeeting to work on systems with older glibc's and
> > > > non-glibc systems too. These systems usually cannot do character set
> > > > recoding on the fly and hence the po files have to be in the character set
> > > > we specify, which is UTF-8. I thought this had been explained umpteen
> > > > times on gnome-i18n@gnome.org.
> > >
> > > Well, we have discussed it, and I think the conclusion was that there
> > > was no need to require utf-8. At least this was the conclusion on the
> > > Gnu translation project list.
> > 
> > That must have been it, because I for sure haven't heard any such
> > conclusion on gnome-i18n@gnome.org. On the contrary.
> 
> So I cannot remember seeing those conclusions... Mybe you could
> refres my memory?

You mean conclusions on gnome-i18n@gnome.org? Sure.
http://mail.gnome.org/archives/gnome-i18n/2002-April/msg00144.html
http://mail.gnome.org/archives/gnome-i18n/2002-April/msg00151.html
http://mail.gnome.org/archives/gnome-i18n/2002-April/msg00153.html
http://mail.gnome.org/archives/gnome-i18n/2002-April/msg00170.html


> > > There is no problem for older systems.
> > 
> > I would love to hear the details on why. I'm sure others would love to
> > hear the reasons too.
> 
> OK, I will dig it out of my mail archives.
> As far as I remember the reason was that they could use the newer
> gettext, as there was no licencing problems - it is under LGPL.
> And this could then be distributed with the distro.

Interesting. But *is* this distributed with Solaris and *BSD and other
systems? Does it work currently?


> > > > > utf8 is a pain in the a**, at least to me..
> > > >
> > > > What's the problem? Please be more specific. I've had no trouble ever when
> > > > using iconv or msgconv.
> > >
> > > utf-8 is just one more source of errors that it would be nice to be
> > > without. I have explained this a couple of times, also on this list.
> > 
> > I'm curious about what type of errors you are referring to.
> 
> It is people trying to edit utf-8 files in latin-1 and then
> the files end up being a mix of utf8 and latin-1 and it is a
> pain in the a** to clear this up. I have had several instances
> like this, mostly with kde files, where they have used utf8
> for some time now.

I think this problem isn't specific to UTF-8 but affects all accidental
mixes of encodings. I think this problem will go away as more and more
editors support UTF-8. Currently "pico" is the only editor I have ever
used that doesn't understand UTF-8.

Also, I'm fairly sure we require that people check their po files for
basic errors with msgfmt before committing them to cvs (this also
catches illegal characters and hence accidental mixes of UTF-8 and other
encodings). Anyone not doing this check shouldn't commit po files in the
first place.


Christian