Build problems?

Ralf Corsepius rc040203 at freenet.de
Mon May 11 17:23:25 UTC 2009


Michael Schwendt wrote:
> On Mon, 11 May 2009 17:49:46 +0200, Ralf wrote:
> 
>> Kevin Kofler wrote:
>>> Michael Schwendt wrote:
>>>> Just to acknowledge that this is not only theory, it has happened before
>>>> actually.
>>> I know, I'm also speaking from experience there. A few made it through to
>>> stable (*cough* rkward *cough* Qt 4.5 *cough* and I think there were a
>>> couple more Qt 4.5 builds which slipped through too early), I managed to
>>> intercept a few others before chaos happened.
>> Then you're more lucky then I am. I am trying to push a chain of 
>> packages updates/upgrades, but I am stuck in a cobweb of broken, 
>> inconsistent repos and mock setups, which allows me to push untested 
>> packages, but doesn't allow me to build them locally.
> 
> What exactly is the problem?

It's all f*cked up:

* Fedora's default mock configurations don't reflect reality:
- They don't know anything about fc12 (n/a)
- FC10's mock builds i386 rpms instead of i586 for Fc11-
- mock doesn't handle rpm's md5/sha256 check-sums correctly (All versions).

* koji/bodji allow building fc11 and fc12 packages


=> I am able to build and push packages into fc11 and fc12, I an mot 
able to build locally, using packages of unknown origin.

> Do you need buildroot overrides as done by releng?
That's a rel-eng's centric view.

My view: rel-reg shouled finally finally start using the infrastructure 
they dictated to contributors. I have fc11 packages pending, I marked as 
stable => they should release them and let "build overrides" die.

Whether as "updates" or as "build-overides" is irrelevant.

> Have you activated any already?
Buildroot overrides? No, because
a) I am actively boycotting them.
b) these updates are non critical.

> Can you build your updates in koji?
Yes, some of them even have been pushed and are in FC9/Fc10 and FC12, 
but not in FC11!

> (you can even submit src.rpm
> to koji as scratch-builds)
> 
> How about opening a new thread about such problems?
No, I feel sufficiently peed, I am not interested and do not have any 
time left to waste on this matter, but prefer users and rel-eng to 
experience the consequences of the brokenness of Fedora's work-flow, 
themselves - These issues are one cause of the flood of (low quality) 
updates users will be facing very short after the release.





More information about the fedora-devel-list mailing list