[Paraview] Fwd: plotting additional variables in paraview

[David Thompson]

Hi all,

I have committed a fix for this issue and verified that ParaView can
read VStress from brick_with_hole2-out.e . I've cleared the continuous
dashboard and will watch the nightlies. It should be in ParaView 3.8.

Berk, please let me know if/when the contract w/ EDF goes through and I
can have the reader tag array components with names.

	David

> While I'm not deep enough into Paraview to contribute specifics to this
> discussion, I appreciate your continuing to pursue a resolution.
> 
> However, as a software developer of analysis codes, I always appreciated
> having names associated with various columns of output data to avoid any
> confusion by the end user (or application programmer, like me) about the
> order of data, especially tensor components.  Or, heaven forbid, having
> the user specify a column number...
> 
> Thanks again,
> John
> 
> 
> On 2/12/2010 9:02 AM, Thompson, David C wrote:
> > Berk,
> >
> > I'll see if perhaps I can bend the new glommer to behave
> > like the previous one until component names appear in
> > ParaView... a few months out sounds like post-3.8 and I
> > wouldn't want to have an official release that misbehaved.
> > But I'm really excited about being able to name components,
> > even though it seems like a small detail -- I guess it's a peeve
> > of mine. :-)
> >
> >      David
> > ________________________________________
> > From: Berk Geveci [berk.geveci at kitware.com]
> > Sent: Friday, February 12, 2010 06:54
> > To: Thompson, David C; Utkarsh Ayachit
> > Cc: John Schiermeier
> > Subject: Re: Fwd: [Paraview] plotting additional variables in paraview
> >
> > Now for the question. One thing I have noticed is that with the box.ex2
> > file in ParaViewData, only part of the stress tensor (sigxx, sigyy,
> > sigzz, sigxy, sigyz, sigzx) is getting glommed because the last 3
> > entries are not in the "canonical" order for a symmetric tensor. I
> > haven't had time to see how ParaView currently handles this but I'd like
> > it to permute the variables into canonical order and glom them into one
> > "sig" variable. The problem is that analysts expecting the order to
> > match the simulation output order will be mislead. What would be nice is
> > if I could *name* each component of a vector/tensor vtkDataArray and
> > have that show up in the toolbar instead of the {Magnitude, 0, 1, 2, 3,
> > 4, 5} combo-box entries that currently appear. It would be easy to do
> > store the names in vtkInformation objects but I don't know how easy it
> > would be to make the ParaView GUI use them (or have VTK copy them down
> > the pipeline). Any ideas? If this is too hard, I will just match the old
> > reader behavior. Otherwise I'll improve the reader's handling of
> > glomming.
> >
> > Funny that you should mention this. We have a deliverable to EDF that is supposed to do exactly this. We haven't thought about how to implement it yet - in the GUI vs. in the pipeline. Do you think it is OK to have a temporary solution until we add this feature for EDF? I can't remember the deliverable date but it is a few months out.
> >
> > -berk
> >
>