darcs no longer removes the _darcs/lock file when ctrl-C is hit while recording
(or doing just about anything else).  I could track this down, but I'd really
prefer to have a test for it first.  Alas, I've spent an hour or so trying to
construct such a test, but have failed.  :( Tricks like

sleep 10 | darcs record
kill %darcs

don't seem to work except interactively--which is not the point, it's easy
enough to reproduce this bug interactively.  Of course, this regression may
depend on stdin/out being a terminal, in which case it may not be possible to
check in the test suite.  :(

David

Interesting. I have reported this previously but now cannot reproduce this
anymore on my repos. Which repo version have you been using for this test? An
old one or a hashed repo? Maybe the difference is there?

On Wed, Nov 05, 2008 at 08:12:01PM -0000, Christian Kellermann wrote:
> Interesting. I have reported this previously but now cannot reproduce this
> anymore on my repos. Which repo version have you been using for this test? An
> old one or a hashed repo? Maybe the difference is there?

The darcs repo.  Perhaps you're using haskeline? Or maybe it's the version
of ghc involved? It's broken for me when compiling darcs with either ghc
6.8.2 or ghc 6.6.
-- 
David Roundy
http://www.darcs.net

I am still unable to provoke this. I have seen a mail containing a patch for the
cabal build system but since we still have Makefiles we need to adress this
issue as well. Is this still an existing problem for you David?

On Fri, Nov 14, 2008 at 06:27:05PM -0000, Christian Kellermann wrote:
> I am still unable to provoke this. I have seen a mail containing a patch for the
> cabal build system but since we still have Makefiles we need to adress this
> issue as well. Is this still an existing problem for you David?

No, my patch to franchise fixed the problem for me, and there never was an
issue with the makefile.

David

Resolved, as pointed out by David and Christian.