Re: gpdf string freeze breakages

[Christian Rose <menthos gnome org>]

tis 2004-05-04 klockan 19.28 skrev Mikael Hallendal:
> On mån, 2004-05-03 at 20:30 +0200, Christian Rose wrote:
> > There have recently been two unannounced string freeze breakages in the
> > string frozen gpdf gnome-2-6 branch.
> 
> Do we really need these public announcement everytime someone makes a
> mistake?
> 
> You said the same thing in the bugzilla and Remi replied 30 minutes
> after you wrote it, give the maintainers a bit of a time to correct the
> mistakes. Like this you have to go apologize in two different places
> just to not look bad.
> 
> Please Christian, give the maintainers some time to correct the problem,
> try to mail them personally first (or in this case, give them at least
> some time to respond). 

I understand your view, but fail to see how it would work in practice.

I haven't always reported freeze breakages to d-d-l. In the past I only
sent them to the offender, the translator list, and the release team.
Why the offender was informed is obvious. The translators were informed
because they need to know why suddenly new unannounced messages appeared
in the frozen release, and the release team because they want to know
about freeze breakages. Noone else was informed.

What happened though is that the amount of unannounced changes and
breakages kept increasing, up until the point that release team members
actually recommended me to make the freeze breakage announcements more
public, in the hope that this would remind developers of the freeze, and
reverse the trend.

So that's what I did. On February 24th I sent a mail to d-d-l
(http://mail.gnome.org/archives/desktop-devel-list/2004-February/msg00516.html)
explaining that all offenders of the string freeze would also be exposed
by name in public on d-d-l from that time on, and that's also what I've
done. It's not something I enjoy doing, it's entirely the other way
around, but fortunately it seems that this at least has had the desired
effect: the freeze breakages appear much less common now.


> I personally don't care about every string freeze that happens in CVS
> since I'm neither on the release team or a translator. And if it was me
> I'd not appreciate (as I know others don't either) to be hanged out in
> the public because of a single mistake, at least not before I had it
> pointed out to me in person and given the chance to correct the mistake.

Hopefully everyone that follows d-d-l should be aware that breaking the
freeze means the danger of being exposed by name. This was clearly
stated in the above mail.
In the cases where it's completely obvious that it's a simple mistake
I've been trying to find the responsible person on IRC first, and only
if that failed sent the breakage announcement mail. However finding
someone on IRC at any moment isn't always possible.

I'm open for suggestions on how to handle freeze breakages better.
However keeping the announcements of them private to the offender, the
translators and the release team doesn't work in the long run. We've
already tried that.


Christian