Re: Belarussian Latin translation

[danilo gnome org (Danilo Šegan)]

Hi chpe,

Today at 14:05, Christian Persch wrote:

> Recently, the "Belarusian Latin" translation team has started adding its
> translations to many GNOME modules. However, they chose to use the
> "be latin" name for their PO files, instead of following the precedent
>>from sr Latn and uz Latn to use be Latn 

Unfortunately, we'll be switching to 'sr latin' instead.  The same is
probably true for 'uz Latn', since they had the same problem with
getting their locale named uz_UZ Latn past Ulrich Drepper and into glibc
(who claimed that the GNU libc established practice was to use
lowercase fully-spelled-out English words for modifiers, whereas there
were only a couple of modifiers at the time, such as "euro" and
"cyrillic": nothing I'd call a rule myself especially since "cyrillic"
was only used with sr_YU cyrillic which was going away, but anyway)

So, even if Latn is an ISO 15924 identifier for Latin script, it made
no difference, so while sr_CS Latn never made it into GNU libc (after
sitting in Bugzilla for ~3 years), sr_RS latin made it in just a
couple of days ago  (when Ulrich responded for the first time to
request to add sr_CS Latn, we already had all of GNOME translated, and
even Serbian KDE team started using it for Latin translations).

Afaik, Fedora, SuSE, Debian, Ubuntu have been shipping sr_CS Latn
locale already, but I don't know about other distributions.

> Can we please resolve this before the GNOME 2.18 release, as to not
> create a backward compatibility problem for our users by having to
> change it in a later release?

Since SVN allows file renames, this is not as big problem as we had
with that until recently.  As far as compatibility is concerned,
people will be able to use both anyhow, and I guess Belarusian guys
are actually better off with be latin since that's what'll get into
GNU libc (yeah, big distributions will pick up anything people
actually use, like they picked up sr Latn, but smaller ones usually
use only what GNU libc provides).

And fwiw, GNU libc has a fallback mechanism, so even if locale with
modifier is not present, the one without is used (so, if we later
rename be latin to be Latn, people will only have to change their LC_*
settings to point to it, and it will display proper translations, but
things coming from locale such as dates will be in the wrong script).

> (This question was already asked on gtk-devel-list
> [
> http://mail.gnome.org/archives/gtk-devel-list/2007-March/msg00044.html
> ] but apparently nobody from the be Latn translation team has
> responded.)

I don't know if Uzbekistani team is aware that they'll also need to
change theirs to uz cyrillic (according to
http://sources.redhat.com/ml/libc-alpha/2003-09/msg00091.html thread,
and especially Ulrich's response).

As far as Serbian is concerned, I wouldn't bother with it right now
since we've got really fast recode-latin-sr script in GNU gettext 0.15
and later, and I want to generate sr latin po on the fly in the future
(so, I'll modify makefiles for GNU gettext, intltool and
gnome-doc-utils to do that).

Cheers,
Danilo