[ubuntu-x] [Fwd: Releasing Mesa 7.6 final for Karmic (was Re: Last changes to X in Karmic)]

[Bryce Harrington]

Tasks that were identified in the offlist discussion.

----- Forwarded message from Rick Spencer <rick.spencer at canonical.com> -----

Date: Wed, 07 Oct 2009 10:41:30 -0700
From: Rick Spencer <rick.spencer at canonical.com>
To: Martin Pitt <martin.pitt at ubuntu.com>
Cc: Alexander Sack <asac at canonical.com>,
	Alberto Milone <alberto.milone at canonical.com>,
	Bryce Harrington <bryce at canonical.com>,
	Sebastien Bacher <seb128 at canonical.com>,
	steve.langasek at canonical.com,
	Araceli Pulido <ara.pulido at canonical.com>,
	Marjo Mercado <marjo.mercado at canonical.com>
Subject: Releasing Mesa 7.6 final for Karmic (was Re: Last changes to X in
	Karmic)

So we need to flip into action mode now. Please:

1. bryce: get this ready to roll out
2. bryce: define the rollback process, along with a time estimate
3. rickspencer3: layout key dates for checkpoints and go/no go decision
4. ara: find internal folks with a variety of hardware, and track their
experience
5. ara: monitor bug reports closely after roll out
6. rickspencer3 and all: define measure for triggering roll back
7. marjo, make final go/no decision by specified time, or before

Please join #ubuntu-x for coordination.

Thanks.

Cheers, Rick

On Wed, 2009-10-07 at 17:00 +0200, Martin Pitt wrote:
> Hello all,
> 
> Alexander Sack [2009-10-07 16:34 +0200]:
> >  1. get all the bits in the archive _today_
> >  2. get QA testing on it right away - as suggested by pitti
> >  3. first review on friday for regression bugs filed after this upgrade; eventually
> >     fix things over weekend
> >  4. review not fixed regression bugs a second time on late Tue/early Wed and decide
> >     if we want to go for RC freeze with this.
> >  5. final decision on Friday after RC freeze - after this its point of no return
> 
> I discussed that with Rick an hour ago, and we came up at a similar
> compromise. 
> 
> My biggest concern, like Alex has as well, is the psychological
> barrier of not wanting to rollback on regressions, and instead
> investing a lot of time and energy of trying to fix the problem
> instead. If we commit to rolling back at the first non-trivial
> regression, then let's go ahead.
> 
> My other concern is that there doesn't seem to be a final 2.6.0
> package, just a 2.6.1+gitsomething. If we upgrade to that, then we
> have the very same problem again: a git snapshot instead of a stable
> release. But if there's a tested 2.6.0 somewhere, let's get that
> uploaded (it was pre-approved in LP #420803 anyway).
> 
> Martin
> 


----- End forwarded message -----