Re: New translators workflow
it would be nice if you could behave in a way so that I don't have the
feeling that I don't exist in the dpkg landscape for you.
FWIW (to the lurkers here) I approved Sébastien on the alioth project, I
recruited him as a French translator, and I informed Guillem about the
fact that I approved him on the alioth project by private mail.
On Wed, 29 Oct 2014, Guillem Jover wrote:
> As mentioned in [T], I'd rather have new translators submit .po files
> to the BTS, as this makes it easier to coordinate changes, handle
> releases, git workflows and similar. I've thus removed your user from
> the alioth project, but certainly not because your contributions are
> not welcome, they are very much appreciated! It is just easier to
> maintain this way.
This is a bit silly. I don't see why it's easier. I actually spent quite
some time to find someone who is interested enough to follow dpkg updates
closely... I don't know if he subscribed to debian-dpkg but we can
certainly ensure that's the case instead of dropping his commit rights.
And I don't understand the double standard compared to other active
translators. Either you impose a standard workflow for all translators or
you let people pick what they prefer...
> I'm wondering too if translators might find it easier to work from
> something like a weblate instance, which then I could pull from time
> to time?
My experience with Weblate (for the debian-handbook) is that it generates
lots of commits even when in mode "lazy commit" and that it will pollute
I do manually squash commits before merging but as soon as you have
multiple translators active (and that happen often with a translator + a
reviewer) you have a stream of commits from multiple people and you
can no longer squash the commits if you want to respect the attribution
of the changes.
The other problem is that as soon as people use such a tool, they
are much less likely to test build their translations and the
rate of syntax failures in manual pages will increase (weblate has feature
to validate XML in strings, but I don't know of anything for
manual pages...). Furthermore the "lazy commit" option means that changes
are not immediately committed and thus a git clone from the weblate
instance will not give you the latest changes (in case you want to test
That said from the translators point of view, it's certainly a good
Raphaël Hertzog ◈ Debian Developer
Support Debian LTS: http://www.freexian.com/services/debian-lts.html
Learn to master Debian: http://debian-handbook.info/get/