Testing libsatsolver on Fedora

[Rahul Sundaram]

On 07/31/2009 06:29 PM, Paul W. Frields wrote:

> James has written about these before:
> 
> http://illiterat.livejournal.com/5043.html
> http://illiterat.livejournal.com/5218.html

Yes, I have read through them before but doesn't quite answer the
question on why comparing two commands that do the same thing is wrong
especially now that sat solver does support file lists and seems to
behave pretty much the same. It seems to a very natural thing for users
to do. I did the same when testing the solv command as well. It seemed
slow in the earlier versions that were posted and I confirmed it was by
comparing a similar operation with yum and now it has improved and I
demonstrated that to myself, again by comparing it with yum.

If others are doing it wrong, then there  *is* a clear and direct way to
demonstrate performance. Yum developers can do the performance
benchmarking between different dep resolves (yum, apt-rpm, smart and now
solv) and publish it in the way that is considered fair. Obviously,
yum's performance and general behaviour (for example, yum search) has
improved drastically but it would be useful to get a good comparison
between these programs.

Rahul