Jan Nieuwenhuizen <[EMAIL PROTECTED]> writes:
> > . at least submit a pot file short before release date.
> That'll only work once the initial translations have been made,
> I guess.
You also have to account (or more precisely, _not_ account) for the
fact that some teams are fast, while some teams are slow. If you wait
after all teams for making software releases, you will become frustrated.
Best is that maintainers and teams work a bit independently. The teams
are responsible for the fact translations exist, once the POT file is
available. Your way to collaborate is accepting the translations which the
robot forwards to you, without otherwise pushing or pulling on translators.
For fast teams willing to follow, it seems that one or two weeks between the
last pretest and the official release is usually enough. For slower teams,
do not try waiting. Things will come in the proper time for those teams.
Just don't take that burden on your shoulders.
> Well that's the biggest problem, imo. Messages, warnings and errors have
> not been really standardised; it's a bit of a mess. I'd like to devise
> a bit of a standard, and have this cleaned up before we do the initial
> submit for translation. Are there any directions available that could
> be useful for LilyPond?
In my own experience as a maintainer, and despite I'm quite sensitive to
internationalisation, translators may pinpoint many inconsistencies before
I see them myself. Of course, you may try to do good, but do not let this
become heavy on you alone. A few iterations, with the translators around,
should sort out most of the things. We are also trying to setup a special
team (the "English" team) meant to address consistency issues in the original
untranslated strings, this is still in gestation -- Karl Berry Cc:ed.
P.S. - One thing the robot does not handle itself is the submission of
POT file by maintainers. I screen out these manually (sometimes not very
fast), and then, the robot takes care of them when I ask it to do so.
The only missing thing is some authentification mechanism, it would be
too dangerous if forged POT files were handled automatically. I'll come
to such authentification devices, and then, things should fly faster. :-)
--
Fran�ois Pinard http://www.iro.umontreal.ca/~pinard