On Fri, Aug 24, 2007 at 08:17:32PM +0200, Eric Y. Kow wrote:
> > Fri Aug 24 09:57:47 PDT 2007  David Roundy <[EMAIL PROTECTED]>
> >   * always run the unit test
> 
> Well... ok.  I guess even if we still have spurious (?) errors, it's no
> big deal because the test doesn't exit with an error.

I've been thinking about disabling yet more of the spurious quickcheck
tests.  In particular, the conflict resolution being valid.

> Speaking of which, look what I just got.
> 
> Checking that conflict resolution is valid... Falsifiable, after 61 tests:
> 
> {
> merger 0.0 (
> addfile ./world/test
> rmdir ./world
> )
> hunk ./test 4
> -5cIy-
> +
> }
> 
> move ./E~I1 ./world/3Bth/x

This is spurious again.  The merger describes a conflict between an addfile
and a rmdir, which means that the directory must have been empty.  So the
move into a subidirectory (3Bth) can't be possible... i.e. this is a bogus
sequence of patches.  But it's just too hard to enforce the constraints in
the Arbitrary instance.   :(
-- 
David Roundy
http://www.darcs.net

Attachment: signature.asc
Description: Digital signature

_______________________________________________
darcs-devel mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-devel

Reply via email to