KDevelop and upstream frustration

[Scott Kitterman]

On Friday, November 23, 2012 02:59:52 AM Aleix Pol wrote:
> Hi,
> Some days ago I had a chat on IRC regarding the current policy of putting
> bugfix releases in a repository waiting to be blessed for entrance in the
> repositories that end-users use.
> 
> I'm unsure how it works in general, but from my point of view of KDevelop
> maintainer it's getting frustrating. We had an unfortunate problem that
> didn't hit the Kubuntu package and the fixed package hasn't entered yet
> (after 1 month?), even though I've been told that it's packaged somewhere,
> so AFAIU the problem is that it's there to be tested and there's nobody
> with enough authority to push that change.
> 
> What I'd like to ask with this mail is what should I do to have enough
> authority for having the latest KDevelop patch release in the main
> repository. I understand that what you need is testing, so maybe I need to
> use a different repository (although that doesn't sound very good to me,
> since then I won't be sure what version are the end-users using) or maybe I
> need some kind of badge, but I think Kubuntu users deserve to have the bugs
> we spend our time fixing as soon as possible and I'd like to help for that
> to happen.
> 
> Cheers!
> Aleix
> 
> PS: I guess that this will be the case for the most upstream but they
> probably just don't know.

The update was released for Kubuntu 12.10 a few hours ago.

As a general rule, we don't update to new versions post release.  There are 
some packages that have exceptions to this policy based on upstreams 
committing to be very strict  about limiting there micro version releases to 
strictly bug fixes (I gather from this and your previous commentary on IRC that 
KDevelop does this).  KDE SC has such an exception.  To get an exception for 
KDevelop would require someone to ask the Ubuntu technical board to give it 
one as well.  

https://wiki.ubuntu.com/StableReleaseUpdates/MicroReleaseExceptions

I would recommend finding someone in the Kubuntu community who is willing to 
work on this for you and work with them.  We can use this update that we just 
did as an example as a successful trial run after waiting a bit to make sure 
no regressions are reported.

I understand it's frustrating.  There is a balance between being to slow and 
going so fast we end up breaking user's systems.  We lean towards caution (in 
the general process) because we place a high value on avoiding post release 
regressions (even if that means it take a little longer to get some things 
fixed).

Scott K