PPA support policy

[Ralph Janke]

Well, generally looked at this issue it seems to make sense. However, I think 
some other issues play in line with this.

1) I personally running pre-4.3 on jaunty because of all the issues and 
regressions in 4.2. IMHO 4.2 is not a stable version on jaunty. Therefore, I 
ended up upgrading due to reports and hope that 4.3 would solve it (well some 
has been solved, some has gotten worse).

I am not sure how many others have done the same for the same reason. 

2) It is also IMHO a big problem that it is not realistic to downgrade back 
(at least I know of no such way that would be possible to be used safely). 
Therefore, it is somehow a catch-22.

However, I usually report most of those problems directly upstream with the 
bughelper. This does comply with the suggested policy, as I understand it. 
However, it is not 100% clear how it can be easily figured out if certain fixes 
are available in Kubuntu packages.

Concluding, I think it is a good idea to encourage reporting upstream 
especially for those problems. However, as long as there are severe issues 
with the stable version(s) (which I consider to be regressions), we may need a 
simple way that allow to figure out when Kubuntu packages are available that fix 
such.

I hope there is a way, this can be done without too much additional work.

Best wishes,
Ralph

On July 5, 2009 02:20:00 pm Harald Sitter wrote:
> Hullos,
>
> As of recently we get quite some reports using KDE pre-4.3 on jaunty
> from PPA. Of which some appear to cause a lot of investigation work.
>
> Some bugs can cause a triage overhead with the result that the issue
> was caused by a bug fixed 3 days later in SVN... So I'd very much like
> to have a policy on what kind of support we provide for PPA
> installations
>
> IMHO: installation-support-only unless
> package-targeting-backports-causes-regression
>
> Because: Until KDE reached .0 it is a fast moving code base and issues
> should be perused upstream, since we mostly provided the packages to
> help upstream aggregate testing anyway. Once it reached .0 we usually
> consider getting the packages into -backports, so we should try to
> minimize regressions and in general ensure it is not making your
> system explode. Still, unless an issues is proven to not affect latest
> stable (e.g. stock jaunty, which would make it a regression) it is of
> no particular interest to us.
>
> The package-targeting-backports-causes-regression policy would of
> course affect stuff like "my KDE ain't starting anymore"... that is
> obviously a regression ;-)
>
> The ultimate triage answer to reports failing policy #1:
> "Please report upstream"
>
> The ultimate triage question to reports failing policy #2:
> "Did this also appear before upgrading to KDE 4.3.0?"
> if no => regression => pursue issue
>
> Issues reported about the development series are always to be pursued.
>
> Establishing such policies will most likely cause bad reactions,
> unless we get standard answers that explain very well why there is
> need for those policies, and of course how to make the report valid by
> reproducing it on stable or development. So, do you think it is worth
> the risk and do you think those policies make sense?
>
> regards,
> Harald