On Thu, Oct 6, 2011 at 2:07 PM, Cody Permann <firstname.lastname@example.org>
I'm running into a serious regression with this patch. If you declare a Mesh variable with the default constructor arguments you end up with a mesh with dimension=1 (mesh_base.h:80). If you then make a call to mesh::read(...) -> ExodusII_IO::read(...), you'll immediately trip the new error check at exodus_io.C:110 when it tests for the dimension of the mesh and fails if it equals 1. The ExodusII interface was updated by Roy some time ago to set the dimension of the mesh "after" reading the ExodusII header information. This new error check simply needs to be postponed until after the header is read and the mesh dimension has been set.
Ah crap. The problem is that the mesh dimension doesn't appear to get set until the very end, in particular following the sideset read that trips up the 1D read. I'm looking into a workaround. Go ahead and remove the error check for now - sorry.