Again: Oh no, something goes wrong

[drago01]

On Thu, Sep 8, 2011 at 9:27 PM, Adam Williamson <awilliam at redhat.com> wrote:
> On Thu, 2011-09-08 at 14:12 +0200, drago01 wrote:
>> On Wed, Sep 7, 2011 at 11:52 PM, Michael Schwendt <mschwendt at gmail.com> wrote:
>> > On Wed, 07 Sep 2011 14:42:07 -0700, AW (Adam) wrote:
>>
>> > Meanwhile, more updates have arrived, but could only be installed
>> > in two steps due to remaining broken deps:
>>
>> I don't really want to bring this up again but the fact that updates
>> with broken deps can end up on the mirros is a bug in bodhi and/or the
>> compose tools. PERIOD.
>
> I'd argue it's not really a 'bug' in the case of updates-testing. You
> can't stop people submitting updates with dependency issues.

It is not about submitting stuff but actual pushing to the repo/mirrors.

> Would it be nice if AutoQA depcheck was 100% accurate and trusted enough
> by maintainers that we could have it block packages that fail from
> making it into updates-testing? Sure. But I'm not sure it's right to
> characterize that as a 'bug'. Kinda semantics, I know.

1) Is it a problem for the users? --- Yes they cannot update
2) Can it be fixed so that it does not happen? --- Yes by doing an
auto depcheck and throw it out before pushing
3) Is 2) simple to implement? Seems not but that does not change that
the current situation is broken i.e a bug