Run of repoclosure against core-development and extras-development

[Panu Matilainen]

On Thu, 21 Jul 2005, Jeff Spaleta wrote:

> On 7/21/05, Panu Matilainen <pmatilai at laiskiainen.org> wrote:
>> What's more worrisome than extras-devel having broken dependencies is that
>> we have them in fc4-extras as well (at least on x86_64):
>
> you have to be a little descriminating and double-check that the dep
> problem you are seeing is from the "newest" package in the tree.  For
> example
> the newest version of octave-devel  in extras is 2.1.71-12, which
> doesn't have a problem.

Ah, ok, I didn't check for that at all.

> repomanage doesn't yet have a facility to ignore "old" packages in a
> tree. And i am very much against making Extras a fully rolling tree
> like rawhide and I believe we need to keep old versions of packages in
> a tree around "long enough" to make sure the new update doesn't have a
> horrible problem so people can revert back to the older package
> without having to rely on a cached copy.

I'm still of the opinion that extras should have a separate "release tree" 
and updates (+ updates-testing) tree just like FC has, even if the 
"release" of FE for some version doesn't happen the very same date as FC 
does. Rolling trees make me ... nervous. With a "release tree" you can be 
sure there's at least one known version of a given software with known 
behavior, buggy or not. With rolling trees you never know.

> If repomanage could magically gain the ability to ignore old versions
> of a package in a tree, this would clean up several of the dep errors
> you saw.. like the octave-devel error and the screem error.

Yup, that'd be a nice addition to repoclosure. I'll try to remember that 
the next time I'm bored and looking for something to hack on :)

 	- Panu -