[openstack-dev] [nova] stable branches & failure to handle review backlog

Thierry Carrez thierry at openstack.org
Thu Jul 31 16:23:48 UTC 2014


Jeremy Stanley wrote:
> On 2014-07-31 10:17:16 +0200 (+0200), Thierry Carrez wrote:
>> That's a good idea. We would probably switch to $PROJECT-stable-maint
>> teams then (each including $PROJECT-core and the general stable-maint
>> team) since we don't have a group in Gerrit for *-core anyway.
> [...]
> 
> I think we can actually do this with the groups we already have,
> just by changing the exclusive group permissions to only be for
> workflow -1..+1 rather than also code review -2..+2? That way the
> stable branches start inheriting the code review permissions from
> the parent refs/* blocks in per-project ACLs. We could try it out
> after the current point release freeze wraps up, if there's some
> consensus.

Sounds promising. If I understand correctly, that would give core people
+2 rights on their project stable/*, but keep APRV rights to stable-maint...

That said, after having spent the day on stable reviews, I confirm that
in most cases reviewers that are not specifically on stable-maint tend
to review the correctness of the patch (like for a master branch review)
rather than only review its applicability to the stable branch (which is
what matters here).

So it looks like we would have to train a lot of people to the delicate
art of stable branch review before those +2 would actually help us in
the review.

-- 
Thierry Carrez (ttx)



More information about the OpenStack-dev mailing list