[darcs-users] Darcs 2.4 - performance testing (stage 0)

Eric Kow kowey at darcs.net
Tue Dec 29 09:07:58 UTC 2009


On Tue, Dec 29, 2009 at 00:15:08 -0800, Jason Dagit wrote:
> > And by any chance, does it have any knock-on effects of making the
> > performance stuff run in a more realistic timeframe?
> >
> 
> I don't know why it keeps timing out.  It's almost as if the OS is doing
> something weird with sleep modes.  I've disabled all the power saving
> options that I can find.  I've told harddrives not to spin down, etc.

Hmm :-( Anybody else have better luck with this?

> Does darcs-benchmark use my global cache?  Perhaps I should mv ~/.darcs
> ~/.darcs.restoreme and see if it makes any difference.

I think that would be good to try.  I think it'd be worthwhile to have
darcs-benchmark systematically create a fresh cache directory for each
kind of repo it uses.

Benchmarking old fashioned repos
--------------------------------
Also: I've thought about the problem of benchmarking old-fashioned vs.
hashed repositories.  One simple option would to be to just ship
old-fashioned tarballs identical to their hashed counterparts and run
darcs-benchmark on those.  We won't get side-by-side tables showing the
difference between the two but we can just copy and paste columns from
the relevant hashed and old-fashioned tables to get the desired effect.

If this makes sense, Petr, do you think you could distribute such
tarballs for use with darcs-benchmark --get?

Thanks!

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 195 bytes
Desc: not available
URL: <http://lists.osuosl.org/pipermail/darcs-users/attachments/20091229/49e1da17/attachment.pgp>


More information about the darcs-users mailing list