On Wed, Aug 26, 2009 at 14:30:00 +0100, Ian Lynagh wrote: > > I think you can divide the problems and proposed solutions into five > > broad categories. > > I think you left out the most important thing: We need a good way to > make performance regression tests.
Yes, thanks for catching that. As part of his GSoC work, Petr has released darcs-benchmark onto hackage: http://hackage.haskell.org/package/darcs-benchmark So all you have to do is to build a couple of versions of Darcs, say Darcs-2.3 vs the current darcs darcs and tell darcs-benchmark to go. Darcs-benchmark also outputs memory use information. More information here: http://wiki.darcs.net/Development/Benchmarks This is important to me because it means (a) we can start backing our claims with hard data and (b) we can start asking for performance patches to be accompanied with standard benchmark data. Darcs-benchmarks more or less got the automation part down, now we need to make it autonomous: http://bugs.darcs.net/issue1569 that is, running on some server somewhere without any human intervention. -- Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow> PGP Key ID: 08AC04F9
pgpWXVChYEsAO.pgp
Description: PGP signature
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
