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

Eric Kow kowey at darcs.net
Sun Jan 10 22:35:29 UTC 2010

I'm a bit concerned because we've been heaping a lot on Petr's plate
(during exam time too!).

Petr, I suspect you'll agree with me that it's a better use of your
time to focus on knocking out the Darcs 2.4 regressions first?

Sorry for piling all this stuff on!

On Wed, Jan 06, 2010 at 20:44:04 +0100, Petr Rockai wrote:
> Well, darcs-1 is currently not possible, since darcs-benchmark expects
> darcs 2 executables. It shouldn't be hard to fix I guess, I am not sure
> why the restriction is there...

Here's my benchmarking wishlist in order of priority.

1. comparing darcs 2.3.x to darcs 2.4.x

     - not enough benchmarks for local operations
     - it's still hanging for Jason Dagit :-(
     - no comments from Windows (Max?)

2. comparing old-fashioned repos against hashed

3. comparing darcs 1 vs darcs 2 binaries

I think #3 will have to take a backseat for now.  Clearly, we very
much WANT to help Nathan benchmark a darcs 1.0.9 binary versus the
latest darcs 2.4.  But we may not be able to afford it, at least not
without outside patches.

As for #2, that's really meant for the GHC team, and if they can live
without them and rely on their informal testing, we may have to back
off from this goal as well.

As you can see, the Darcs Team is still needs way more hacker time.
If you'd like to help us with benchmarking, please

  darcs get http://repos.mornfall.net/darcs/benchmark

No special skills needed, just a little elementary Haskell.

> I think it would be best to only have a single tarball of any given
> repository and use darcs get --old-fashioned or so to convert it back
> into old-fashioned if needed.

This would be for #2

Perhaps one cheap way to do this would be to build this into
  darcs-benchmark --get
When it fetches a tarball, it could open it up in a playground,
get --old-fashioned and save that up as old-fashioned tarball.

Then darcs-benchmark by itself can be completely ignorant about it.

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/20100110/f90d9fbb/attachment.pgp>

More information about the darcs-users mailing list