On Sun, 11 Oct 2015 10:49:48 +0200, Stephan Kulow wrote:
Am 10.10.2015 um 16:12 schrieb Takashi Iwai:
For a KMP build problem, it's easy to see, though. We can set up an OBS project (e.g. Kernel:stable:KMP) building only KMPs based on the latest Kernel:stable. We have already something similar for SLE, and this has helped to catch build issues beforehand.
That would really helpful
The setup should be trivial. Michal?
(if Kernel devs actually check that project too, but usually the bot submits the new kernel and that one doesn't care).
Well, actually it's more package maintainers who need to work on the breakage by the upstream kernel update. The kernel devs are willing to help, of course, though.
But in this case VB and bbswitch were actually broken (indirectly) by the new systemd submission that ended up in Factory at the same time and the real problem is - as already highlighted elsewhere - the missing coverage for these use cases in our openQA testing. And I'm not exactly sure we want to cover *everything* - otherwise we will spend 2 days testing to publish and snapshot and that can't be the goal either.
Right. Constantly delaying even more than now is really bad for users, too.
So having some gaps in TW that require workarounds - like installing VB from devel project - are fine IMO. We should cover what most people do - and I'm not so sure using VB is one of that.
I wonder whether we may have more lightweight test coverage for some cases. openQA is really great, but it's mostly lifting up the whole heavy system. The kernel team is trying to look for a CI testing, and a similar method might be used for KMPs, too, I suppose; at least loading a module would work in most cases even though it's actually not used. thanks, Takashi -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org