On Mon, May 10, 2010 at 10:06 PM, Lennart Regebro
On Mon, May 10, 2010 at 21:38, Matthew Wilson
wrote: I know how to use timeit and/or profile to measure the current run-time cost of some code.
I want to record the time used by some original implementation, then after I rewrite it, I want to find out if I made stuff faster or slower, and by how much.
Other than me writing down numbers on a piece of paper on my desk, does some tool that does this already exist?
If it doesn't exist, how should I build it?
Since circumstances change, most importantly hardware, the only way to be sure is to run the performance test on all versions on the same machine (obviously while it's not doing anything else). A system that can take a bunch of svn/hg/whatever tags, check them out build them, and test them as a a big batch could be helpful.
you can also translate the results into pystones, so your output is roughly the same on every machine that has a similar architecture / OS I guess.
Don't really see what it has to do with distutils...
I think Matthew also wanted a way to record previous runs and diff them. That would be a good question for the Testing In Python mailing list,
-- Lennart Regebro: Python, Zope, Plone, Grok http://regebro.wordpress.com/ +33 661 58 14 64 _______________________________________________ Distutils-SIG maillist - Distutils-SIG@python.org http://mail.python.org/mailman/listinfo/distutils-sig
-- Tarek Ziadé | http://ziade.org