Re: merging kparal/rpmguard-integration into master

[James Laska]

On Thu, 2010-01-07 at 14:15 -0500, Will Woods wrote:
>  On Thu, 2010-01-07 at 12:01 -0500, Kamil Paral wrote:
>  > Hi,
>  > 
>  > it seems to me that the kparal/rpmguard-integration branch is now
>  > mature enough to be merged into master.
>  > 
>  > You can try the rpmguard test by issuing:
>  > 
>  > 1. checkout kparal/rpmguard-integration branch
>  > 2. $ /usr/share/autoqa/post-koji-build/watch-koji-builds.py --dry-run
>  > 3. pick a line
>  > 4. $ (copy that line) -t rpmguard --local
>  > 
>  > And look to stdout or rpmguard.log. Example rpmguard.log:
>  > http://pastebin.com/f534d362f
>  > http://pastebin.com/f42ca41cd
>  > 
>  > What do you think, any objections to merge? Send all blame to me,
>  > thanks.
>  
>  Everything looks good to me. Awesome work. I'll pull this into master
>  shortly.
>  
>  > PS: It doesn't currently send results by email, same as current rpmlint
>  > test. I hope we will discuss enabling them both soon after.
>  
>  Yeah, to do that we're going to need a library function to look up the
>  owner(s) of a given package from the pkgdb. Shouldn't be too hard,
>  though.
>  
>  What's really interesting (but will be much trickier): I'm pretty sure
>  we're going to want to be able to hold/block/untag packages with certain
>  failures. That will require a whole big policy discussion with the
>  packaging committee and rel-eng, and some way of letting maintainers
>  sign off on expected failures.. 
>  
>  We'll get there eventually. But probably we should start thinking about
>  exactly which problems should cause a package to be rejected, or held
>  for review. Does anyone know of an existing list/policy to use for a
>  starting point? 
I'm not aware of anything so far.  Kamil's work on rpmguard has sparked
a lot of excitement for me on the testing possibilities with package
updates.  I don't know if it's too early to head down this path, but do
folks think the time is right to start working towards a [[QA:Package
Update Test Plan]] to outline the details and move the discussion
forward?

Thanks,
James