Emergency checkin procedures

[David Miller]

David Miller wrote on 10/12/05 3:11 PM:

> If you have a situation where the trunk is broken and it causes dataloss 
> or prevents Bugzilla from being usable, backing out the patch that 
> caused it is not workable due to subsequent checkins or data conversion 
> that's already been done, and you can't find someone to approve the 
> patch to fix it in a timely manner, get two reviewers to agree to it, 
> and then go ahead and check the fix in.  Please leave the approval flag 
> requested on the bug, as it'll still need to be checked for approval 
> retroactively, and make sure to justify in comments in the bug why you 
> needed to proceed with the checkin.
> 
> Please only do this when it really is a critical regression.  Situations 
> where this permission is utilized will get reviewed, and if I find 
> people abusing it, this permission will go away.

I had pointed out to me a few minutes ago that this has always been the 
procedure, and thinking back on it, that's true, it has been.  I've been 
a bit forgetful lately on top of everything else and it slipped my mind 
that we've already been doing this in the past.  It's always been a bit 
of an informal thing though, so consider this getting it down in writing. :)

-- 
Dave Miller                                   http://www.justdave.net/
System Administrator, Mozilla Corporation      http://www.mozilla.com/
Project Leader, Bugzilla Bug Tracking System  http://www.bugzilla.org/