Re: Here we go again *SIGH*

[Vincent Untz <vuntz gnome org>]

Le mercredi 04 mars 2009, à 12:21 +0000, Anna Jonna Armannsdottir a écrit :
> On þri, 2009-03-03 at 14:41 +0100, Vincent Untz wrote:
> > I would also point out that having no freeze break (with requests from
> > the developers, of course) would be bad: having some breaks is a good
> > sign, showing that the development is active and not dead. Of course,
> > we
> > don't want tons of freeze breaks. And of course they should be
> > approved
> > first.
> > 
> > Vincent
> 
> The responses from the developers indicate quite clearly that they 
> are not satisfied with the methods of string freeze. 
> Presently a number of developers, are deliberately breaking the 
> string freeze and thereby creating trouble for translators. 
> The argument seems to be: "We break string freeze because we can". 

I think it's unfair to say developers are using this argument or even
that they're deliberately breaking the string freeze. If you look at the
breaks, we have the following situations:

 + the string was not marked as translatable.
   => we want the break anyway, since it can only make the situation
      better.

 + the developer asked for permission before committing.
   => this was the decision of the coordination team to accept the
      break.

 + the developer didn't ask for permission, but it was approved
   afterwards.
   => this is bad and the developer should be told so. But the result is
      the same as the above case.

 + the developer didn't ask for permission, and it was rejected
   afterwards.
   => this is bad and the developer should be told so. The break gets
      reverted. (if it doesn't get reverted, please tell the release
      team)

 + this is needed for a really really really important bug fix (eg,
   related to security)
   => it doesn't happen often since it's really an exception, but in
      such cases, the opinion of the coordination team is bypassed (but
      the coordination team would likely approve such a change anyway)

 + the module does not follow the GNOME freezes (eg, gtk+, glib)
   => not a lot we can do, except ask the maintainers of such modules to
      think of translators (and I would say they do)

I don't think developers break string freeze because they can. When they
do so before asking for permission, it's mostly because they forget
about it. It'd be interesting to look at what happened this cycle and
see how much of the breaks had no request (and were for completely new
strings, and not to mark existing strings translatable).

Thanks,

Vincent

-- 
Les gens heureux ne sont pas pressés.