On Thu, 2021-01-28 at 15:15 -0800, Kevin Fenzi wrote:
On Thu, Jan 28, 2021 at 03:05:16PM -0800, Troy Dawson wrote:
> As we are getting closer to the F34 branching, which means we are
> getting closer to CentOS 9 Stream, which will eventually be turned
> into RHEL9 Beta, and then RHEL9 release. Now seems like a good
> time
> to get ideas flowing about EPEL9.
>
> I'm just throwing ideas around. Nothing I'm saying here is even
> close
> to policy or a final plan. If people have other ideas, feel free
> to
> say them.
>
> epel8-next is getting closer and closer to being in place.
> To me it seems logical to create a epel9-next, pointing at the
> CentOS
> 9 Stream (when it comes). It would need the same setting up as
> epel8-next, all the steps would be the same other than the name and
> where it points for it's repo.
>
> We could also setup some type of signup board for if maintainers
> want
> the EPEL Packaging SIG to automatically bring their packages over.
>
> With epel9-next in place, and good set of EPEL9 packages in it,
> users
> would be able to test RHEL9 much better in it's beta phase.
>
> Also, it would take alot of pressure off when we start getting
> regular
> EPEL9 setup. If it takes a month or two, people wouldn't be as
> concerned, because they could always just grab the packages from
> epel9-next.
I think that could be workable, but I'll toss out another proposal:
As soon as centos 9 stream exists, we create epel9-playground and
allow
people to branch/add packages to it. Once rhel9 is GA, we setup epel9
as
usual and epel9-next and point epel9-next to build against stream and
playground to build against rhel9.
epel9-playground acting first as a "Rawhide" for c9s pre-RHEL9 GA and
then as a playground for RHEL9 could be a bit confusing?
The advantages of that would be that epel9-playground is more rawhide
like... it would compose every night and there's no bodhi overhead.
Of course to be confusing we could just treat epel9-stream that way
until GA too I suppose.
Right, using epel9-next but with no Bodhi gating until GA seems like a
nice idea. To add another variant to this: we can also start enabling
Bodhi but with time-to-stable set to 3 days (like Fedora betas) once
RHEL 9 is in beta? i.e. "we think c9s should have stabilized enough by
now that we can start gating EPEL packages targeting it".
Best regards,
--
Michel Alexandre Salim
profile:
https://keyoxide.org/michel@michel-slm.name