On Wed, Mar 28, 2007 at 07:45:23AM +0200, Thorsten Leemhuis wrote:
On 27.03.2007 20:13, Axel Thimm wrote:
> On Tue, Mar 27, 2007 at 04:29:05PM +0200, Thorsten Leemhuis wrote:
>> Opinions? Just ignore that? Or simply do a script-build mass rebuild of
>> EPEL5 with increased %release?
> I'd say ignore it. The current users of EPEL are on this list.
I'm not that sure on this -- EPEL got even mentioned quite a bit in the
press already. Those two were the first ones afaics:
That doesn't matter I think. It is not that we argue on being
secretive, we never announced that the repo contained packages that
were already intended for public consumption, the only official source
of information, the wiki, is scattered with warnings on beng work in
progress. The repo was and is of a rawhide state where we are allowed
to do errors and undo them w/o resorting to trickery to get upgrade
paths right. E.g. we may remove, rebuild, change the layout etc. at
will until we decide we're there.
> We'll have also other issues to resolve (like qt4 on EPEL4
> qt4 on
> RHEL5, and if we check we'll probably find more) and we need to keep a
> rawhide attitude (aka flexibility) otherwise we'll have to start using
> epochs before launching.
Regarding that I'd say: Simply remove the current qt4 package from EPEL4
and try to get one in with a lower EVR then the package in RHEL5.
Yes, and you can only do that if you assume that the state of the repo
was tagged as "experimental, no upgrade paths ensured".
And we may hit more, we haven't yet QA scripts in place, we don't even
have done any manual QA, and we argue that we value testing and QA
even more than Fedora.
So let's keep the experimenal-rawhide-like tag on while gathering
contributors and building/testing the set.
Axel.Thimm at ATrpms.net