Re: Transifex instance for GNOME

[Petr Kovar <pmkovar gnome org>]

Hi all!

"Johannes Schmid" <jhs jsschmid de>, Fri, 15 Oct 2010 10:06:39 +0200 (CEST):

> Hi!
> 
> > I'm absolutely not opposed to such an idea. But I'm also not convinced
> > that we will be able to keep current D-L functionalities. However if we
> > get some new ones, maybe the global balance might still be positive.
> > I think the main objective would be to support (a) auto-commit in GNOME
> > git infrastructure (the blocker is currently on sysadmin side) and (b)
> > auto-commit in an i18n branch of each tracked branch for modules not
> > hosted in GNOME infrastructure.
> 
> I personally think that it would simply disburden us from lots of
> development work we had to put into damned-lies. Not that damned-lies is a
> bad system, actually it is pretty good (thanks Claude!) but as we are the
> only ones using it we also have the sole responsibility.

On the other hand, we are the only ones who decide on what goes in and what
goes out as far as features are concerned. There is some advantage in that
fact. :-)

Anyway, for a change, I'm opposed to the idea of moving to Transifex (or
any other new infrastructure, for what it's worth). It's nothing against
Transifex per se, really, but I'm simply afraid that this move wouldn't be a
win-win situation at the end.

Among other things, we'd need to:

* learn and explain to others what this change means to admins,
coordinators, and our large group of translators,

* sort out issues on different, less centric-oriented workflow
approach in Transifex, this approach and its (dis)advantages have been
discussed many times on this list, as you all know,

* as has been already noted, find people who are actually going to work on
this (I haven't seen any volunteers in this thread as of yet, but I might
have overlooked someone).

The last point is not so trivial as it might seem: from time to time I
follow discussions at Fedora l10n's mailing list, for some time they've
used quite an old Transifex version at translate.fedoraproject.org with
some major glitches:

http://lists.fedoraproject.org/pipermail/trans/2010-September/007950.html

The thing is, now we have stable Damned Lies at GNOME, but in case we
decide to move, ultimately we could end up at the same or similarly
unfortunate situation, just before the important GNOME 3 release. More than
lack of manpower to do the necessary infrastructure updates, we need
stability and working environment, I guess.

> > Here are some steps we could follow to go further:
> > * Find a team of people to take the job
> > * Install a Transifex instance on GNOME infrastructure
> > * Configure Transifex for GNOME needs
> > * Test Transifex global functionality (pilot languages?)
> > * Decide to switch (or not to switch)
> 
> I have seen several people working on/with transifex on this list in the
> past. I hope they can give it a go.
> 
> We need some kind of timeline though. We need to have everything setup and
> working latest on string-freeze day. And we also need an implememtation of
> our string-freeze-break-detection for transiflex at this point.

Well, it's clear that, whatever product we end up using, we'll need to
implement the missing features (or better said, find someone to do it). I
don't see any reason why the missing bits couldn't be implemented in the
existing infrastructure. If nothing else, this would save us a lot of time
and possible troubles from having to do a migration to a completely new
l10n system.

My 50 hellers,
Petr Kovar