Catalyst v5.9 issues

I’ve been doing more testing of the Catalyst changes and wanted to report what I’ve found:

  • It looks like the VTPD writer/reader is now properly writing out and reading in field data and this is being used to deal with channel information in the new API and script generation. This fixed some issues I noticed before so thanks for doing that.
  • The VTPD reader doesn’t allow selecting which fields to be read in. I’d like to see if with the Catalyst V2 API and scripts if we can select specific fields to be passed in through the adaptor (i.e. the adaptor only produces the requested fields selected for the VTPD reader) and used when generating scripts.
  • The Catalyst/CxxImageDataExample is segfaulting. I did a separate build (not in the main ParaView build) and used the script to test. I also tried the attached script (1.9 KB) after changing the code to use the V2 pipeline. It looks like with the V2 pipeline is always assuming that it’s a V1 Python script. Am I supposed to zip up the Python script myself if it’s V2 script? It looks like the segfault is happening upon exit.

Note that my testing is with commit 566212127a0c6ee1e972ec0d48e7a9300ce1a2c8

Also, maybe a comment should be put in the generated Python code that registrationName is the Catalyst channel name for those that are used to previous jargon and may get confused what registrationName is.

That’s the way the examples are currently coded (ref: !4322). I’ve updated the examples in mr !4565 to use API provided to detect Python script version automatically. Thus, there won’t be any need to create a zip unless convenient for other reasons.

If I use my MR here and run the Catalyst/CxxImageDataExample as follows, it seems to works without any issues and produces datasets/*.pvti as expected.

I’ve reported a feature request here.