Re: Planned changes to D-I "level 1" translation framework
On Di, 2007-04-10 at 14:38 +0200, Christian Perrier wrote:
> Quoting Jens Seidel (email@example.com):
> > On Tue, Apr 10, 2007 at 01:31:29PM +0200, Frans Pop wrote:
> > > However, as long as it is very clear to everybody that this split is only
> > > to set a _priority_ for what to translate first, the current split would
> > > be OK with me. It should _not_ be an excuse to stop translating once
> > > level 1 is finished and activation of languages should only be done on
> > > the understanding that *all* of d-i needs to be translated.
> > Considering this I think it's useless to split the PO file. Right?
> > Only inofficial builds containing partial translations would profit
> > from the split.
> My intent was not triggering a discussion about the language
> activation criteria now. But I guess we won't escape from it.
> That discussion should happen and a mistake we made for etch was not
> having it before the release was getting closer.
> My personal opinion is that we should activate a language when
> sublevel 1 is complete....while the final D-I release should only keep
> those languages that have *both* sublevels activated.
> This is more or less what we did for etch....even if the criteria for
> activating languages was not really precise.
> Activating early allows for more testing (until someone figures out
> how to make daily builds from packages built from the SVN and not from
> uploaded packages).
> One problem we have with new languages is the lack of real testing,
> because translations only reach the arhive when the packages are
> rebuilt, which only happens when a beta/RC release is planned.
> We really should try to improve this for lenny.
Forgive me for not searching the archives, but earlier there was talk of
a proper gettext based implementation that allows fallback in case the
translations are not 100%. I know the concern was that not everybody
wants English fallback, but was there any progress on this or veto