> > OK. Knowing that it used to work is already a god thing. Do you happen to > have a vague idea of the last time it worked? >
Just use Aug 1st and test. Who knows, maybe there has always been a bug in the code? I haven't had time to test your example at all because I'm too busy with research at the moment. Time for me to learn how to use hg dissect... > It's `hg bisect` and here is the faq that deals with using bisection on PETSc: http://www.mcs.anl.gov/petsc/documentation/faq.html#bisect -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120118/5d195b85/attachment.html>