On Mon, 1 Jun 2009, Eric Kow wrote:
On Sun, May 31, 2009 at 21:16:14 +0100, Ganesh Sittampalam wrote:
It particularly needs testing on a corrupt hashed-format repo, and on
Windows.
Attached is a list of hashed repositories on code.haskell.org (including
darcs 2 repos)
Would a giant tarball of all the repositories on code.haskell.org be a
good thing here? I have a 400M tarball in my cho home directory
(~kowey/obliterated-20.tgz) which contains repos that we did the timing
tests on way back when c.h.o upgraded to darcs 2. It's each repo minus
20 patches, though and I don't think I can find the original.
Thanks, I'll take a look at those repos.
One thing to note is that some users have very large global caches, so
darcs gzcrcs on any one repo could be very slow. I wonder if recycling
the --no-cache option would be useful for real users, or if it's only
relevant to folks like me who are purposefully trying to test this out.
Would it be useful maybe to advise moving the global cache out of the
way until the user is ready to cope with it and after it's been
checked/repaired once.
I think --no-cache would be a very good idea. I'll do that.
Ganesh
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users
