Yes, that should work. But you need to create at least one visualization module in mayave, e.g. Surface (this can be done also from the command line: mayavi2 -d cylinder.vtk -m Surface), and set the data source to correct field.


Works without any problems.
 
As for the original problem - do you use WX widgets for both mayavi and matplotlib? If so, try setting (just guessing) the matplotlib backend to something else, as sfepy might import matplotlib somewhere during postproc.py call.

I changed the matplolib backend to qt4agg according to <http://matplotlib.org/users/customizing.html#customizing-matplotlib>, I get the same error.

Regards
jagan