-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
thus Josh Boyer spake:
On Wed, Jan 5, 2011 at 7:45 AM, Karanbir Singh mail-lists@karan.org wrote:
Can you quantify what you mean by 'open approach' ( basically, what steps and what gains those steps would bring about )
The aim was to focus people's attention to the upstream beta, better product and that loop etc. We could have started earlier, sure. But now that we have started 2 months back and your own contribution status stays at nil, why are you interested ?
How and what should be contributed if the normal user didn't even know what problems remains?
Problems remain where ? in CentOS or RHEL ? It was RHEL6 that had a public beta, for issues that should have been reported against bugzilla.r.c; or am I misunderstanding what you said ?
Don't get me wrong, I am well aware of the fact that there are issues and situations that need looking at and changing. But lets do the right thing rather than just doing something. Going by the popularist current mood of people on this list, I think people just want early access to a codebase they can start using for their own use rather than actually working towards building CentOS-6. Which makes me fear that the only way we are going to get C6 out of the door in the next few weeks is by clamping up, talking to the usual-suspects and just going back to the CentOS-5 process. And to be honest, I don't really think these conversations over the past two months have been wasted; but in the grant scheme of things - getting 6.0 out of the door might be a better target for now - as long as we can somehow agree that we get back to this process engineering immediately after so as to not be in the same situation, come 6.1.
Also, failback to the CentOS-5 process isn't necessarily a bad thing - we know it works :)
As a fairly new subscriber, I've not really found anything technically wrong with the process. Mostly because I have no idea what the process is. From what I can tell, the CentOS developers pick up SRPMS, debrand, "magically" build them somehow, and then publish them when done.
There probably isn't anything wrong at all with that, but since it's not documented anywhere and the buildsystem being used isn't documented either, it's sort of a big black box. So to me, a more "open" process could start with simply documenting the actual process, including the buildsystem and/or build order of the packages.
Something like Fedora's koji would be nice:
http://koji.fedoraproject.org/koji/index
Maybe I've missed this somewhere in the wiki already, but if I have it's because it's hard to find (at least in my looking).
FWIW, I rather liked the call-for-help on debranding, even if it had limited participation. I think it's exactly the kind of task that can be distributed to relative new-comers and might lead to further participation.
josh
Timo