I just pushed a couple of fixes to PCFIELDSPLIT in petsc-3.3 (
http://petsc.cs.iit.edu/petsc/releases/petsc-3.3/rev/943dde820f7f,
http://petsc.cs.iit.edu/petsc/releases/petsc-3.3/rev/5a9ebf885615)  In
particular, the second changeset
modifies the way the splits' nullspaces are configured.  In my view the old
behavior was not quite correct.
Here are the relevant comments:
http://petsc.cs.iit.edu/petsc/releases/petsc-3.3/rev/5a9ebf885615#l1.9
and
http://petsc.cs.iit.edu/petsc/releases/petsc-3.3/rev/5a9ebf885615#l1.37

However, I don't have use cases where the old way of handling block
nullspaces
leads to a failure.  If there is justifiable legitimate usage for the old
ways, the
changeset can be easily backed out.

Dmitry.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120819/e19aad22/attachment.html>

Reply via email to