Note, this output is generated by maybench <http://code.google.com/p/maybench>,
specifically, its darcs-benchmark program.  It works fine with the yi repository.

Running... /usr/local/bin/darcs get --quiet /foo/xmonad main
Running... /usr/local/bin/darcs changes --quiet --repodir main
unrecord last 1
Running... /usr/local/bin/darcs unrecord --quiet --last 1 --all --repodir main
0.10443s
record
Running... /usr/local/bin/darcs record --quiet --all -m "test patch" --repodir main
0.014715s
obliterate last 1
Running... /usr/local/bin/darcs obliterate --quiet --last 1 --all --repodir main
output:
  |
error:
  |
  |darcs failed:  Couldn't undo patch in working dir.
  |user error (Error applying hunk to file ./XMonad/Core.hs)
  |
darcs-benchmark: user error (Error when running command (/usr/local/bin/darcs
obliterate --quiet --last 1 --all --repodir main))

On Mon, Mar 03, 2008 at 03:36:35PM -0000, Eric Kow wrote:
> Note, this output is generated by maybench <http://code.google.com/p/maybench>,
> specifically, its darcs-benchmark program.  It works fine with the yi repository.

Perhaps you could write a shell script that would reproduce this error?

David

Seems to also fail on TagSoup.  Attached.

Attachments

• bencherror

I've reproduced this now.  Thanks!

David

Okay, I've tracked this down as a time stamp error.  If we add --ignore-times to
record and obliterate, it disappears.  I'm sure we could also achieve the same
with a sleep before the unrecord.  This is old darcs behavior, so I'm not sure
it counts as a bug here.

David

I should add that it *would* be nice to make darcs more resistant to
accidentally-identical timestamps.  This is worse in darcs2, because we do not
also check file lengths (because it would be more expensive in hashed
repositories, since we don't store them--but we could change that).

> Okay, I've tracked this down as a time stamp error.  If we add --ignore-times to
> record and obliterate, it disappears.  I'm sure we could also achieve the same
> with a sleep before the unrecord.  This is old darcs behavior, so I'm not sure
> it counts as a bug here.

Is --ignore-times really worth it in practice?

On Tue, Mar 04, 2008 at 16:02:00 +0000, Eric Y. Kow wrote:
> Is --ignore-times really worth it in practice?

(or rather, the non-ignoring of times)

On Tue, Mar 04, 2008 at 04:08:01PM -0000, Eric Kow wrote:
> On Tue, Mar 04, 2008 at 16:02:00 +0000, Eric Y. Kow wrote:
> > Is --ignore-times really worth it in practice?
> 
> (or rather, the non-ignoring of times)

Yeah, I had it disabled for darcs-2 and Simon M. complained about darcs
whatsnew taking a minute or more.  That's not really acceptable.
-- 
David Roundy
Department of Physics
Oregon State University

I believe this is fixed by the change I recently made which only allows
timestamp comparisons when we've also got file lengths to go on.  This makes
timestamps much safer to use.  And with the sha2 hashes we also are now storing
file lengths.

David