micro release exception for LibreOffice

[Martin Pitt]

Kees Cook [2012-06-13 14:18 -0700]:
> It was pointed out to me that this "get some history of good SRUs"
> isn't actually part of the documented process:
> https://wiki.ubuntu.com/StableReleaseUpdates/MicroReleaseExceptions
> 
> Should we explicitly mention this?

I think it would be a good idea to add it, yes.

> Otherwise it seems like this isn't really what MREs were for. I.e.
> an MRE is being requested to expressly bypass the SRU review
> process, not because it has a strong update history in Ubuntu, but
> because it has strong upstream quality controls.

At least the converse should be true, though: We should not grant an
MRE if the recent history with new upstream microreleases has shown
regressions which slipped through QA. But I could not in good faith
give a MRE for a package which never went through a few successful
upstream microrelease SRUs to demonstrate that the process is working.

I do realize that this is somewhat a chicken-egg problem of SRUs vs.
MRE. If the SRU team insists, we could introduce an intermediate
"probationary MRE" state which is valid only once or twice, and then
make it permanent if it succeeded. But I think it should suffice to
point out to the SRU team that a MRE has been requested by the TB and
now needs to go through an actual SRU. This is the case for the
current LibO update.

Martin
-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)