Plotting is broken with --enable-complex

[libmesh]

Plotting is broken with --enable-complex

[dknez]

I'd like to fix this at least for Exodus. If someone could point me in the right direction, that'd be helpful. I recall that @pbauman had some thoughts about how to unify the codepath so that IO works properly for --enable-complex across the board?

[pbauman]

Yikes. I wish I'd just done this when I was thinking about it. Sorry.

The path I'd had in mind was to mimic what I did for vector-valued variables: in both EquationSystems::build_variable_names and EquationSystems::build_solution_vector (search for TYPE_VECTOR), check if variable is complex. If so, add variable names for each component (I'd had in mind mimicking what was done in GMV output) and plopping the corresponding solution values in the correct index locations for each component. This should Just Work (TM) for Exodus (and I think all the others too except VTK), but GMVIO would have to be updated to use build_solutions_names/vector.

I may have 1. forgotten something and/or 2. Not thought it through all the way, but I think it should be simple and only a few extra lines. Don't have time to contribute code on this ATM, but let me know if the code looks nebulous in there and I'll do my best to shine some light.

[dknez]

OK, thanks. I'll try to get around to this soon. For now I just wanted to sort out the "game plan". What you describe above sounds good, and I'll let you know if I have any other questions once I get around to working on this properly.

Thanks!

[dknez]

This should be resolved in pull request #195 from Dana Christen.