[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: debconf template translation

On Fri, Feb 28, 2003 at 01:52:21PM +0100, Tim Dijkstra wrote:

> > How about you continue filing bugs with translated templates for the
> > time being? The release is not anywhere on the horizon.
> You understand that's almost an infinite amount of work? (Although it
> can be automated.)

What kind of idiotic phrase is "almost infinite"?  There is a finite
number of packages with a finite number of debconf templates, which
undergo a finite number of revisions.

> > If by the time sarge seems to shape up a lot of debian packages lack
> > translated templates, I guess every maintainer should expect NMUs of
> > their packages in order to get them in (not right now, but before or
> > at the beginning of the freeze)

> This means that we'll have to do that before every freeze, and it will
> need some (extra) human intervention on every case. If we are serious
> about internationalization we _need_ a system that circumvents the
> package maintainers. They shouldn't need (and want) to have anything to
> do with translations to languages they not even vaguely understand.

If you are serious about internationalization, you _need_ to develop a
system that brokers effective communication between translators and
maintainers.  Any system that excludes maintainers from this process is
doomed to result in out-of-date translations, because translators won't
even begin working on the templates until the package has been uploaded.
This alone makes it suboptimal, even if (best case scenario) it only
affects users who track unstable.

C.f. the GNU translation project, which has been very effective at
localizing software *with* the cooperation of authors.

> In the future we will have (want) to translate not only debconf
> templates, but also manpages, README's, etc. These are almost never
> supplied by upstream for all languages.

Nevertheless, they BELONG upstream.  That means this work should be
coordinated with upstream, not shoved into a separate package somewhere.
The maintainer is quite often the best liaison between Debian work and
upstream, and this should not be overlooked.

FFS, why are so many people eager to create an infrastructure where the
very translators who are most qualified to bridge gaps between
national language communities are *discouraged* from communicating and
collaborating with package maintainers?  Package maintainers SHOULD be
bothered with (not necessarily bothered by) updated translations; they're
a valuable reminder that our work is useful to a diverse body of users in
a global community.

> If I counted correctly there are now 15 languages using the ddts
> system, which of course are not ready at the same time. This would mean
> that package maintainers are overly bothered by new versions of
> translations all the time. Of course one could wait until everything is
> translated, but that would be a shame for the users (of the testing or
> sid) that want to use languages that are translated already.

I think it's possible to strike a happy medium here between shotgun bug
reports and delayed access to translations.  I think a key component is
getting templates into the hands of translators BEFORE the package is
uploaded -- which obviously requires the involvement of maintainers.
But even beyond that, I think packages whose maintainers are on the ball
will have translations encorporated relatively quickly even after the
first upload containing the template; and packages whose maintainers are
MIA will always need special attention, so splitting translations out of
the package does nothing to address the fundamental problems that still
need to be dealt with.

Steve Langasek
postmodern programmer

Attachment: pgp2mMizS5AC2.pgp
Description: PGP signature

Reply to: