Changes to the SRU processes?

[Scott Kitterman]

On Tuesday, July 17, 2012 01:03:43 PM Brian Murray wrote:
> On Tue, Jul 17, 2012 at 03:59:28PM -0400, Stéphane Graber wrote:
> > On 07/17/2012 03:53 PM, Sebastien Bacher wrote:
> > > Hey everyone,
> > > 
> > > We got some recent SRU issues where regressions were not detected before
> > > having the upload copied to -updated. Those issues were discussed with
> > > the people who maintain the packages and it an obvious flaw in both
> > > cases is that the maintainers are just too busy to keep up with incoming
> > > bugs and didn't spot the new reports due to that.
> > > 
> > > Since that's likely to happen again I would like to discuss some
> > > improvements to the current process:
> > > 
> > > - while I don't like to add extra steps in the way of SRU updates,
> > > should we discuss having a "check bugs reported since the SRU was
> > > uploaded for any potential regression" on the list of steps before
> > > copying a SRU to -updates?
> > > 
> > > Not sure that should be the SRU team responsability, maybe the uploader
> > > should do that when the upload is flagged "good to go"?
> > > 
> > > That's probably not needed for every upload so what about using a new
> > > tag for the updates that seem worth that extra check before being
> > > copied?
> > > 
> > > - some people asked what we could do to improve the chances that
> > > reported regressions are flagged as such. Do we currently tag the bugs
> > > concerning -proposed or -updates versions in some way? If not, could we
> > > do that? It would narrow the list of bugs to check out for potential
> > > regressions
> > > 
> > > - other ideas of what we could improve...?
> > 
> > Can't we have a bot spotting new bugs matching the version in -proposed
> > and marking verification-failed the master bug in such case (or any bug
> > linked to the SRU, so that it's blocked)?
> 
> That makes more sense than just tagging some random bug as I'd
> suggested.  I'd also add a comment to the verification-failed bug
> linking to the bug using the -proposed package.  Given this idea a bot
> sounds like a better idea than modifying aport.

This sounds to me like something that could have caught the recent problems, 
so +1 from me.  Double +1 since it's a solution that doesn't result in more 
random bug mail for me.

Scott K