Why was a kernel-2.6.34 pushed to updates that had un-addressed bugs.

[Richard Ryniker]

>In my case (F13, x86_64 on a Lenovo X200) the .34 kernel made suspend fail. 
>On laptops I think working suspend/resume should be blocker for release. It 
>worked before, hence it is a regression.

F13 was released with a .33 kernel, therefore the question of blocking
the F13 release for this reason did not arise.  I presume you question
whether F13's update to kernel .34 should have ocurred.

Was the suspend problem recognized before the .34 kernel release?  Did
this kernel fix other issues that might be judged more important than
suspend/resume?  Did suspend fail for all equipment?  Rare
configurations?  A different set of machines (yours and others that used
to work, failed; different models that used to fail now worked)?  Maybe
Bugzilla has some data, but it might not be a simple issue.

In any case, fallback to an earlier, installed kernel that works is
easy.  Without the wider testing and QA process that goes into a
numbered Fedora release, I am not surprised an incremental kernel broke
something.

With kernels, and most other packages, I think there is some presumption
that upstream testing and development has combined elements in a way that
makes sense, that balances benefit against risk, that may mix faults with
fixes to achieve a net positive effect.  Sometimes this simply is not
true, but I doubt the person upstream who decides to release an update
believes it is false.

It would be nice to not have regressions, but an effort such as Fedora
that seeks to deliver quickly the latest software technology cannot, in
my opinion, avoid them.  If regressions or other faults occur too often,
the protocol for distribution of Fedora updates might be improved.
Consensus about a precise definition of "too often" may be difficult.
This interest list often contains laments from users that update X, which
fixes their problem, is not available in a Fedora repository, often
because the update has not acquired sufficient positive test results.