I just ran the tests with 2.0.4 of Gmsh on OS X and everything passes
(other than one test case
unrelated to gmsh). I also did the steps on <http://matdl-osi.org/
fipy/wiki/SurfaceOfSphere> and
everything worked as expected.
Charless, If you take the steps on <http://matdl-osi.org/fipy/wiki/
SurfaceOfSphere> with trunk
and gmsh 2.0.x, do you still get a patchy surface?
Which version of pyvtk are you using? I'm using 0.4.67.
Cheers
On Mar 20, 2007, at 1:17 PM, Jonathan Guyer wrote:
On Mar 19, 2007, at 6:45 PM, Charless Fowlkes wrote:
Perhaps backface culling is somehow turned off in your vtk setup
so that you can't see the mis-ordered faces?
That could be it. We got gmsh 2.0 importing working yesterday, and
I also tested doing a manual 1.0 export from gmsh 2.04. Neither
exhibits the patchyness that you see.
I don't know, offhand, how vtk was compiled (and dread getting into
it), but we clearly need investigate.
On reflection, I realized that for your application, solution is
probably fine, despite the bad face ordering. The inner and outer
faces are automatically no-flux, regardless of which way they are
pointing. The radial faces are the ones that matter for the physics
and they should be OK.
I still haven't been able to track down where in MayaviViewer the
vertex list for each face is getting incorrectly re-ordered.
My gut reaction is that the problem is the orientations as imported
into the mesh, not how they're being passed to Mayavi, but thanks
for your diagnostics. We'll definitely use them to help track down
the problem.
--
Daniel Wheeler
