Package EVR problems in Fedora 2007-10-31

[Ralf Corsepius]

On Fri, 2007-11-02 at 10:53 -0400, Jesse Keating wrote:
> On Fri, 02 Nov 2007 15:48:36 +0100
> Ralf Corsepius <rc040203 at freenet.de> wrote:
> 
> > No disagreement on this. But I really don't see what you gain by
> > forcing EVRs to "updates < testing < rawhide" for "packages in
> > testing".
> 
> That's simply because you're failing to consider that things which are
> in testing have (high) potential to make it into stable.
That's "updates < testing" ... a necessary condition, because otherwise
you won't be able to install a package from "testing".

>   It would be
> better for maintainers to fix nevra issues while the build is still in
> testing than to wait until it hits updates.
Are you saying packages in "testing" automatically hit "updates"?
This would be the next design flaw. This renders "testing" further
useless.

I sense we seemingly we have a basic divergence on the purpose of
testing. 

You seem to understand it as a "delay queue" for updates, giving some
people a chance to check packages and withdraw them when they feel they
need to.

I understand "testing" as "auxiliary repo" taking candidate packages for
"updates", which generally should only be pushed by request, not "by
timeout" nor by "no receiving complaints".


Ralf