[vtkusers] CGNS

[Berk Geveci]

> (warning: long winded reply to simple question) I have been looking into 
> various file formats looking for something standard, useful for 
> unstructured meshes and flexible enough to be used  with a numerical 
> code (eg. finite element/volume). I had looked at the HDF5 Mesh API but 
> it looks like an abandoned project. 

Yeah. It looks it has not been touched for a while.

> Also, a problem with its design is 
> that it doesn't provide a way to define groups of nodes/elements which 
> could be used to specify boundary conditions (eg. how hot a wall is).

I guess this is what happens when a file format is designed by
visualization people :-)

> After a some more searching I found CGNS (who I believe are also 
> planning to introduce HDF5 internally). 

It looks like they are working on it. I think this would be a big
improvement. 

> So far it looks  like what I want and yes, if I can't find a ready made 
> reader for it in my favorite 
> visualization toolkit then I would plan to write one me self, which 
> leads me to your next point...

I am certainly willing to help you.

It looks like I have one problem with CGNS. It does not look like they
support parallel IO nor spatially distributed data.

-Berk