Release-critical bug process?

[Adam Williamson]

On Wed, 2009-02-11 at 14:04 -0800, Jesse Keating wrote:

> What's been done so far as grown organically mostly due to 2 things.
> 
> 1) We're huge.  Really huge.  Fantastically huge.  Waiting for all our
> software to be perfect to do a release would mean we'd have releases
> even less often than Debian.
> 
> 2) Everybody has a pet bug, that they will consider mission critical.
> That's just the nature of the game.
> 
> At the end of the day, there had and has to be a dictator, a BOFH,
> somebody with the gumption and the authority to just flatly say "No".
> Either "No, we're not going to ship without this being fixed." or "No,
> we're not going to delay the release just to fix this issue."
> 
> For better or worse, that person has basically been me since I started
> doing releng for Fedora.  I rarely if ever make any such decision
> without consulting many parties first, but at the end of the day, I'm
> responsible for seeing that Fedora gets out, and accountable that it
> goes out in a usable manner.  I consult many people and try to inform
> even more, but waiting for a vote of some sort from a (large?) body just
> doesn't scale IMHO.
> 
> I'm all for adding more transparency to the process that evaluates the
> potential blockers and makes the call on go/no-go per bug.  But that
> process should be pretty light weight and capable of being done quite
> quickly, particularly because "blocker or not" tends to be pretty
> subjective depending on a lot of factors, that just might not be seen by
> the BugZapper alone.

Just for the record, a few people discussed this further on IRC, and
basically it seems the current process is probably the best. Useful
points that arose:

a) We should see the point of alpha releases as being basically to
provide an updated installer tree. Only bugs that break the installer or
prevent you being able to get logged in and run 'yum' should be
nominated for the Alpha release trackers.

b) We all more or less agreed that the process should make it quite easy
to nominate a bug to 'block' a release, and bugzappers will probably do
a lot of that, but ultimately for now it's probably right for Jesse (or,
more formally, RelEng) to be the ultimate arbiter. Jesse would like us
to err on the side of nominating things to be blockers - he's more
worried about missing something important than the time it takes to weed
out things he doesn't consider critical.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org
http://www.happyassassin.net