[Paraview] CTH Reader/Parts extract problem

[Rick Angelini(CISD/CIV)]

I have a CTH 8.1 dataset that I'm loading up using a development version 
of Paraview.     When I load up the data and use the "Extract CTH Parts" 
filter, everything seems to work fine with the exception of some 
unexpected holes in the resulting geometry.       Due to the shape of 
the holes, I'd say that it looks like Paraview is not collecting all of 
the extracted features from the servers.      I have a build from the 
CVS repository from April that works fine - but the builds I have from 
August CVS downloads have the missing geometry/holes.     

So, at this point, I don't know if it a problem with the CTH reader, the 
"Extract CTH Parts" filter, or some oddity of OpenMPI that I built against.

Just to verify that the data itself is OK, I've done a couple of tests:
- if I load up the data running only 1 pvserver, the geometry looks 
fine, although quite slow to load!  8-)
- if I create a slice through the original geometry on the at the same 
location as one of the holes, the resulting clip plane is correct.

Any recent changes to either the CTH reader or CTH part extract function 
that might explain this?