On Thu, Jun 21, 2012 at 2:39 AM, Robert <libmesh@robertweidlich.de> wrote:

At least VTK supports also vector-valued solutions of arbitrary size.

It looks like the current VTKIO bypasses the build_solution_vector paradigm and does its own thing. Which means the splitting of variables won't apply to VTKIO. On the other hand, it probably means something will have to be done explicitly for the VTK case to handle the vector-valued variables. I've got an vector_fe example that I'm using to work out the (numerous...) kinks ATM. Once that's going, perhaps you could use that to help add that support? I know nothing about VTK internals and, since we use ExodusIO and GMV around here, I'm afraid I don't have the time to implement that capability.