[Libreoffice] [ANNOUNCE] Branch libreoffice-3-5-0 created

Petr Mladek pmladek at suse.cz
Fri Jan 27 05:27:17 PST 2012


Jan Holesovsky píše v Pá 27. 01. 2012 v 11:54 +0100:
> Hi Stephan,
> 
> On 2012-01-26 at 18:13 +0100, Stephan Bergmann wrote:
> 
> > > 	+ only cherry-picking from libreoffice-3-5 branch
> > > 	+ 2 additional reviews needed; 2nd reviewer pushes
> > 
> > Is that to be interpreted as, once a commit has reached libreoffice-3-5 
> > (where it already needed original committer plus one reviewer), it does 
> > need another *two* additional reviewers (i.e., original committer plus 
> > three reviewers total) for libreoffice-3-5-0, or is one additional 
> > reviewer sufficient (i.e., original committer plus two reviewers total)?
> 
> I guess Petr meant here the former - ie. after having pushed to
> libreoffice-3-5, two additional reviewers are needed to be able to push
> to libreoffice-3-4-x; it is the same as in -3-4.

Exactly, 3 reviews are currently needed for the 3-5-0 branch. It is
intended for the last release candidate. It will be marked as final few
days after it is available => there is almost no time for testing => we
should be very careful about changes.

The idea behind these x-y-z branches is that testers requested at least
10 days for testing of the last release candidate. It is plenty of time,
so why not doing one more release candidate in the meantime with
super-safe changes (translations, trivial fixes, super reviewed blocker
fixes). Such changes would make happy developers and users and will not
trigger extensive testing.

Heh, I am getting really scared by the flood of patches waiting for
3-5-0 approval. Some are not trivial and it is getting dangerous.


> But if you think / feel that actually one additional would be enough, I
> guess we can talk about that at one of the next TSC's?

Sure, we could discuss this. I think that we do not need 3 reviews for
translation update, dictionaries update and similar things that are
usually safe and can't be reasonably checked by developers.

On the other hand I vote to keep 3 approvals even for one liner fixes.
Any change might have strange effects and there is not enough time to
found them with the last rc.


Best Regards,
Petr



More information about the LibreOffice mailing list