Alsa-utils update broken

[Ralf Corsepius]

On 02/15/2012 06:00 PM, Michael Schwendt wrote:
> On Wed, 15 Feb 2012 13:10:00 +0100, RC (Ralf) wrote:
>
>> [...] Playing down the issues and to wipe them under the carpet
>> doesn't help anybody.
>
> Who's "playing down the issues"?  Don't make up such things
Well, as I read your response, you have been accusing the packager, 
trying to shift respnsibilty away from rel-eng.

Have you ever heard the word "final inspection" (Germ. "Endkontrolle")?
In Fedora this is rel-eng => this kind of breakages is rel-eng's 
responsibility.

> These issues are a topic not just on this mailing-list, but in (all?)
> Fedora related community support channels. Clearly they are not hidden
> under the carpet. That's enough bad publicity, and 3rd party repos
> add additional broken deps. Still one doesn't need to post sarcastic
> comments or misinterpret/misunderstand the reason for the broken dependency.
> It's simply wrong to conclude that testers should have noticed it.
>
>> Fact is: After all these years Fedora is around, rel-eng is still
>> pushing packages with broken deps, despite QA and AutoQA, and Fedora's
>> bureaucracy.
>
> And one thing still hasn't changed either: if this is your pet peeve
> issue, why haven't you done anything in all these years that extends the
> existing infrastructure with the missing feature?
<sigh/> I am can't get involved into everything.

>> I am not blaming the alsa-utils/libs packager(s), I am blaming those
>> people who are supposed to assure the distros' releases and updates are
>> consistent.
>
> I only added that pushing inter-dependent updates as multiple bodhi
> tickets makes it harder (if not impossible) for existing testers to catch
> broken deps like this.

rel-eng had 8 years to think about this problem and to develop 
appropriate strategies. They obviously didn't.

> The testers run with updates-testing enabled and
> don't see any "yum update" failures, because everything needed is
> found. It would be insane to request them to check low-level package deps
> manually. If inter-package dependencies are so strict and only the
> package maintainer knows the order in which to push individual packages
> (which are supposed to be backwards compatible!), karma automatism in bodhi
> could have been turnt off. No need to rush. The full ALSA stuff could have
> spent much more time in updates-testing before pushing it manually. Also,
> if testers have tested alsa-utils *with* the needed new alsa-lib, why let
> push alsa-utils without the needed alsa-lib?