Procedure to push a package causing broken deps to f15?

Marcela Maslanova mmaslano at redhat.com
Wed Feb 16 16:07:34 UTC 2011



----- Original Message -----
> From: "Ralf Corsepius" <rc040203 at freenet.de>
> To: "Development discussions related to Fedora" <devel at lists.fedoraproject.org>
> Sent: Wednesday, February 16, 2011 12:44:44 PM
> Subject: Re: Procedure to push a package causing broken deps to f15?
> On 02/16/2011 10:55 AM, Kevin Kofler wrote:
> > Marcela Mašláňová wrote:
> >> I completely agree.
> >>
> >> At least nag-mails about the broken dependencies (because of
> >> rpm-4.9)
> >> should be delivered sooner or we should wait with branching or do
> >> mass-rebuild sooner. Now we have to build everything for F-{15,16}
> >> and
> >> even wait for testing,
> The real problem behind all this is these packages having made it into
> f15 - This should not have happened, QA should have caught them
> earlier,
> should have fixed them or at least have informed these packages'
> owners.
> 
Autoqa should be able to track these dependencies, but it's not ready yet.

> >> which is ridiculous in this case. It's only delay
> >> fixing of broken deps.
> Agreed, Note my wording: I call this a "delay queue", not a "QA input
> queue", because it's effectively a mere delay queue without any
> 
> > Well, I proposed a way to fix the procedure:
> > http://lists.fedoraproject.org/pipermail/devel/2011-February/148604.html
> >
> > What do you think of that?
> Well, then the same will happen with the beta freeze.
> 
> To me, the key would be not to let packages causing broken deps into
> the
> repos. QA should preform an analysis on how they made it into the
> repos
> (In my understanding, the cause this time, was an improperly merged
> last-minute mass rebuild of the perl-modules, which due to a change in
> rpm's dep-tracking is causing broken deps).
> 
> That said, I would propose to immediately push package updates for f15
> to testing (spares ca. 24 hours of delay) and to reduce the
> "testing->stable" push delay to 24 hours or less.
> 
> Ralf
> 

I suppose this was already denied by FESCo, but I could be wrong.
We were definitely speaking about shorter periods in testing, but the
period looked to short and updates probably won't be tested at all.

Marcela
-- 
Marcela Mašláňová
BaseOS team Brno


More information about the devel mailing list