[Yum-devel] API Changes: Was: plugings modifying the URLGrabber instance

[seth vidal]

On Tue, 2005-08-02 at 05:06 -0700, Panu Matilainen wrote:
> On Sun, 3 Jul 2005, seth vidal wrote:
> 
> > [SNIP] the earlier discussion of what is modern system maintenance. I'll
> > answer that more later.
> >
> > In the meantime - I think we have some amount of consensus that we want
> > to make the repo/packagesack/packageobject code more functional for
> > different types of data storage.
> > This is a non-trivial amount of conversion work once we get the new
> > classes written. So I'm wondering - do we want to put out a yum 2.3.4,
> > now and then work on these changes. Mainly b/c the amount of changes
> > will be pretty huge, I think, we'll be debugging them for a while, I'm
> > guessing.
> >
> > Thoughts?
> 
> I've been wondering for a while now would it actually make sense to stamp 
> 2.3.4 + fixes in current HEAD as yum 2.4.0 and leave the huge changes for 
> 2.5. Just the enormous speedup in 2.3.x thanks to sqlite would be well 
> worth a new release IMHO (not to mention all the other goodies) and 2.3.4 
> seems very stable to me. As an added bonus putting out 2.4 should shut up 
> most of the "but apt is faster" whiners as well :)
> 
> Thoughts?
> 

It's a fair point. We could stabilize on current HEAD and break compat
here.

Benefits:
 
 - we could spend some time on cleanup on 2.4 for distros like centos4,
fc4 and others
 - speed advantages you've already explained


Problems:
 - we may need to be able to break -HEAD anaconda and pup in this cycle
for fedora and not having to worry about older releases could be in our
benefit.
 - Is everyone comfortable with how things are in 2.3.X right now? Is
there anything glaringly outstanding that is a concern?

-sv