Processes in the L10N team

Thomas Canniot thomas.canniot at laposte.net
Sun Jan 7 09:00:45 UTC 2007


Le dimanche 26 novembre 2006 à 13:54 +0100, Bart Couvreur a écrit :
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Dimitris Glezos schreef:
> > (removed -devel-list since this is a -trans-list issue)
> >
> >
> > O/H Thomas Canniot έγραψε:
> >> Le dimanche 26 novembre 2006 à 01:21 +1000, Chester Cheng a écrit :
> >>> Hi Igor,
> >>>
> >>> The .pot file hasn't been prepared by the author.
> >>> It will be released soon and we can make po files then.
> >> I think there is translation related problems here.
> >>
> >> Why hasn't it been made BEFORE FC6 ?
> >> Why hasn't it be done at the same time than developing desktop-effects ?
> >> Why translation is always seen as second class interest by developers ?
> >> Why most translators seem to accept this second class idea ?
> >> Why bugzilla bug reports must be opened to force a developer to add
> >> translations ?
> >>
> >> Question I don't have any answer for ... yet ?
> >
> > Thomas,
> >
> > I agree we need to fix our l10n process. I've spoken to some people and the
> > presence of the gap is pretty much known. How can we fix it then?
> >
> > Here are some first ideas:
> >
> >   1. Start by documenting the need for attention to this matter, for
> example by
> > getting some numbers of the non-english users (e.g. fedora-brazil is HUGE)
> >
> >   2. List irritations the translators have stated in the past. One
> example is
> > releases not having up-to-date translations in packages [1].
> >
> >   3. Start having IRC meetings to discuss things, our progress and get
> people to
> > hang out on the IRC channel more often.
> >
> >   4. Think about electing a Steering committee for the team; check out the
> > DocsProject voting proposal:
> >
> >     http://fedoraproject.org/wiki/DocsProject/Policy/FDSCoElections
> >
> >   5. Start writing guidelines for developers and slowly try to make
> them happen.
> > A steering committee can help with this by having open communication
> channels.
> >
> >   6. Open up a wiki page holding links to common/known problems, a bug
> tracker
> > for translation bugs (is there one?) and start pushing release-blocker
> bugs for
> > important issues.
> >
> >   7. Get the team closer to the Docs project; this team does a *great*
> job and
> > the two teams have a lot in common and could share experience, tools etc.
> >
> >
> > The L10N project shouts "I need resurrection" with all it's strength.
> So, to
> > have something to hold onto, I copied the above points here:
> >
> >   http://fedoraproject.org/wiki/L10N/Tasks
> >
> > Ideas, comments, suggestions?
> >
> > -d
> >
> >
> > [1]: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=207095
> >
> Nice list Dimitris, I totally agree.
> 
> I would like to add one more thing: the biggest problem (IMHO) is the
> fact that the L10N-project is way outside the Fedora devel-space
> (cvs-wise). Lots of package maintainers know about our efforts here,
> but just don't see it, because of the fact that they have to use two
> cvs-accounts / trees. I know the plan is to get ourselves on
> cvs.fp.org and maybe we should just go for it. So what do we want when
> we do the migration??
> 
> An idea:
> - -> get hooked up in the Fedora Account System, now people on l10n and
> docs need 2 accounts to get work done, cut it down to 1
> - -> get a seperate cvsroot on cvs.fp.org with only .po / .pot -files.
> - -> try to get into Jesse Keating's pungi, for l10n-inclusion (not so
> sure if this is possible, if not just a hook in the build system)
> - -> set up an sort of sponsor-managed cvs or dir in the cvsroot (as
> extras, docs) so only language maintainers can commit (this may mean a
> lot of pain for the maintainer, but gives more control) and those
> files get into the build system

And this definitely improves the quality of the translated string.

> - -> (try to) migrate the web-interface and maybe add an
> online-translation system to it, so we create a low barrier to the
> project and thus more contributors + QA

I completely aggree, a bit late, with Bart.
>From my point of view, I don't think it could be that indispensable to
have the same kind of webapp than Launchpad offers for translation, that
is,translation directly from the web browser. I personally don't trust
any web browser, their stability depending on the websites you are on.

> - -> hook up a commit-mailinglist (or maybe just this one??)

It would be great as well, so as to check for updates.

> - -> ....
> 
> Maybe we could call for a meeting with Fedora Infrastructure and Docs
> when this list is ready and get this "issue" off our back :)
> 
> Bart

-- 
Thomas Canniot
http://fedoraproject.org/wiki/ThomasCanniot




More information about the trans mailing list