AutoQA: distro congestion?

[Michael Schwendt]

On Thu, 21 Apr 2011 10:42:04 +0300, AT wrote:

> I think when it becomes normal to circumvent the system then perhaps we
> need to rethink it and implement it in a sane way?

I wouldn't say "it becomes normal", but certainly there's a tendency
towards wanting updates be pushed into the stable repo _as soon as possible_.
Especially those updates where the package maintainer knows what he's doing.
Also those, where a few guys try to bump the karma level even before an
update has been distributed widely enough by the mirror system. 
The "minimum time to spend in updates-testing" is not put to good use.

Add to that all those updates, which need to get marked stable by their
submitter without anyone having spent karma points on them (and with
bug reporters in bugzilla not commenting on them either).

For those the Update Acceptance Criteria are just a hindrance.

[...]

And with all the testing requirements, major breakage slips through
nevertheless, because we're missing testing instructions for nearly all
updates.

<rant> Some weeks ago someone rushed out a mesa-dri-drivers upgrade
for F-14 that made projectM stop working for ATI/Radeon. The corresponding
assignee at Red Hat doesn't answer in bugzilla. Yesterday evening, an
incoming crash of Audacious on F-14 made me think "WTF?", and this
time it was a bad projectM update that has not been tested
painstakingly despite me having raised a few essential questions
in the bodhi ticket.</rant>