[Paraview] Linking Catalyst example samples with Catalyst

Praveen Narayanan praveenn at nvidia.com
Thu Jun 12 14:30:31 EDT 2014


Not sure if this is the right thread to ask since my original installation difficulties have now resolved themselves. 

A few days ago, miniFE instrumented with catalyst was put up at Andy Bauer's git repo (https://github.com/acbauer/miniFE-2.0_ref_Catalyst), which I built (successful- first make the adaptor and then make the main code) and tried running (unsuccessful). 

From what I see, we should call the python pipeline creator by appending the name of the file gridwriter.py. However, I cannot get it to show any of the data rendered on screen when it does the coprocessing although Catalyst::coprocessor creates vtk data structures both for the grid and the data associated with the grid, and the python script creates the pipeline in the method doCoprocessing (is that the correct interpretation?)

Currently, it connects to catalyst and executes the pipeline since it dumps data files to the disc and generally responds to the simulation. {\bf But I cannot get it to render anything live}.

Thanks
Praveen.




-----Original Message-----
From: Ben Boeckel [mailto:ben.boeckel at kitware.com] 
Sent: Friday, June 06, 2014 7:16 AM
To: Praveen Narayanan
Cc: paraview at paraview.org
Subject: Re: [Paraview] Linking Catalyst example samples with Catalyst

On Fri, Jun 06, 2014 at 02:18:42 -0700, Praveen Narayanan wrote:
> I am examining the example code for catalyst from Andy Bauer’s git 
> repository (https://github.com/acbauer/CatalystExampleCode).
> 
> I am trying to learn about the catalyst workflow before using it in my 
> own in situ examples.

So, first, there is no requirement to use a Catalyst-the-build to use Catalyst-the-coprocessing-API. Catalyst builds are meant for use where a smaller ParaView is a better fit (super computing, lower memory constraints, etc.). The naming is indeed confusing :( .

> python catalyze.py -i Editions/Base -o ../catalyst_src
> 
> fatal: Not a git repository (or any of the parent directories): .git
> 
> Error: Command '['git', 'rev-parse', '--show-toplevel']' returned 
> non-zero exit status 128

You can give `-r <dir>` to tell it the top-level of the source tree.
Without it, it will try and find the top-level using git, which won't work for tarballs. I'll improve the error message.

> 2)  I tried two of the catalyst sources
> 
> a.  Base+python: This builds properly and links with the example code
> (CxxFullExample)
> 
> b.  Base+essentials+extras+python: This also builds and links with the 
> example code, but I get the following runtime error:

When do these errors pop up? The Catalyst build trees are slimmed down ParaView builds and don't have everything under the sun included (e.g., animation classes are missing). If you need other classes, another edition should be made to include the required classes and proxies.

> 3)  The build base+python runs the example code, but does not connect 
> with catalyst. Furthermore, upon running make test to test with the 
> sample python script in ‘SampleScripts’ (which basically invokes 
> CxxFullExample with the python script as an input argument, I get 
> errors stating that it cannot load some vtk modules, which upon 
> looking turn out not to be built at all

<snip>

> I would like to know what I am missing in these builds. It appears 
> that some of the vtk modules are not being built:
> 
> Error: Could not import vtkCommonComputationalGeometryPython
> 
> Error: Could not import vtkRenderingCorePython
> 
> Error: Cannot import vtkPVServerManagerDefaultPython
> 
> Error: Cannot import vtkPVServerManagerRenderingPython
> 
> Error: Cannot import vtkPVAnimationPython

Catalyst builds do not build all ParaView modules, so some missing is expected.

> What is the correct way to get a working run for the examples 
> supplied. Also, how do we connect to catalyst after this?  My 
> understanding is that we just load up paraview (4.1) and then the 
> simulation would connect to catalyst after we hit ‘Connect to 
> Catalyst’ and load up the ‘Coprocessing’ plugin. Does paraview have to be built from source?

The way I've done it is that the simulation loads up Catalyst (using the vtkCPProcessor class). ParaView can create Python scripts for use with vtkCPPythonScriptPipeline, but if there is a way to use ParaView-the-application with in-situ runs, I don't know it. Andy?

> More specifically, is there a particular download version that might 
> work, and what flags do we turn on in CMakeLists.txt? Is there any 
> other set of examples (although I think the git examples demonstrate 
> the workflow quite properly) that we could use to try catalyst?

From a full build, PARAVIEW_ENABLE_CATALYST=ON should be all that is necessary. From a Catalyst build, the same should be sufficient, but for any VTK or ParaView classes which you require, you will need to create an edition to bring them in.

--Ben

-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain
confidential information.  Any unauthorized review, use, disclosure or distribution
is prohibited.  If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------


More information about the ParaView mailing list