speed of yum depsolver

[Jay Cliburn]

seth vidal wrote:
> On Mon, 2007-03-12 at 17:13 +0100, Tomas Janousek wrote:
>> Hi,
>>
>> On Mon, Mar 12, 2007 at 02:25:27PM +0100, dragoran wrote:
>>> yum now uses its own depsolver instead of rpmlib which is slower. I know 
>>> that its planned to make it working first befor starting to optimize it.
>> Is it possible to turn it off and use the old one? It just cancelled it after
>> it had been running for four hours, hoping I'll be able to use the old one,
>> because this is just ridiculous. I wonder how apt-get does it, it's able to
>> give me the list of updated/installed/whatever packages in less than a minute
>> on a 32MB RAM Celeron 233MHz box and yum needs more than 4 hours on 256MB RAM
>> vmware on dualcore Pentium on 3Ghz?
>>
> 
> I'm getting a little annoyed at this conversation.
> 
> Apparently it is not acceptable for things to be not-working or
> less-good in rawhide. Apparently, we have to go from one working state
> to another perfectly working state w/o passing through any situation in
> which it is not as functional.
> 
> We told everyone up front that what is going on right now is slower and
> larger while we work on getting it right and yet that doesn't seem to
> stop the useless bitching and moaning w/o any offers of help at all.

IMHO, this discussion brings into sharp relief just how good Rawhide 
really is.  It's so good that I've become accustomed to near release 
quality software at any and all points in the Rawhide life cycle. 
Occasionally something breaks -- spectacularly, even -- but it often 
gets fixed within a few days.  This yum-in-transition performance issue 
is just another bump in an otherwise pretty darned smooth road, all 
things considered, and I'd argue that if Rawhide were more unstable, the 
instant yum issue wouldn't seem nearly so contemptible.

I like running the latest stuff.  I like Rawhide.  I like Fedora.  You 
guys are doing a great job.

An alternate voice from a satisfied Rawhide user/tester,
Jay