Re: Interesting anaconda/yum performance issue

[Jeffrey Law <law redhat com>]

On Wed, 2007-06-13 at 18:09 -0400, Jeremy Katz wrote:

> > Understood -- and I certainly need to fully specify versioning
> > information and the like (though I don't use globbing).
> 
> I suspected as much :-)
:-)  Basically it gives me the ability to recreate a system, even
if people are adding newer versions of packages to a repo, changing
group membership, etc etc.  As long as packages aren't removed, I'm
a happy man.


> [snip]
> 
> Yep, pretty much that exactly.
Cool.  I'll keep using that patch locally until an update appears.
It should be interesting to time one of the larger installs -- I've
got one profile which burned an additional 20 minutes when it was
converted from groups to explicit packages.




> 
> > I'm not sure what to do with the casematch argument or if the other
> > calls to parsePackages ought to be changed too.  And I'm certainly
> > not well-versed in python, anaconda or yum to know if there are other
> > issues I'm not dealing with.
> 
> Not specifying casematch is definitely the right thing to do.  It could
> well make sense to change other calls to parsePackages.  I'll go ahead
> and get this into upstream yum after we do the yum 3.2.1 release.
I doubt the use in "exclude" is likely to be time critical, nor should
the one in "include" be time critical.  However, if you manage to kill
all uses of parsePackages, then you've got a nice cleanup to go along 
with the performance improvement for the "install" case.

> 
> Thanks for doing the experimenting for me!
Couldn't have done it without a pointer to matchPackageNames -- Thanks!

Jeff