generate_polydata: python script requires vtkpython

The goal of this repository is to store python script to help user to use VTKHDF File Format. Reducing the number of dependency is better for taht.

Currently only this file needs vtkpython, this script should be updated to not depends on it.

It would be very helpful for user not familiar with VTK.

https://discourse.vtk.org/t/trying-to-make-a-temporal-vtkhdf-polydata-using-python/15298

changed title from generate_polydata to generate_polydata: python script requires vtkpython

changed the description

FYI @louis.gombert @mwestphal

well the branch I'm working on to add HTG support uses VTK to generate data and convert the XML format to VTKHDF, so I do not support the idea. I don't see the issue with requiring VTK for a tool that generates data specifically made to be read by VTK

The objective of this repo is NOT to use VTK to generate VTKHDF files.

  # varying number of amrbox between timesteps
  amrbox1_values = [
    # t0
    [0,3,0,5,0,9],
    [6,9,4,9,0,9],
    # t1
    [2,7,2,7,0,9],
    # t2
    [4,5,4,5,0,9]
  ]
  amrbox1 = level1.create_dataset('AMRBox', data=amrbox1_values)

sure, this does not use VTK. But what am I supposed to learn from that someone who wants to implement a writer?

what we need to do on this repo goes much beyond simply removing VTK. Dumping values with no meaning into a HDF5 file does not bring much more documentation value than taking them from a VTK dataset

The objective is not to document the datastructure of VTK, but we may need to rely on the doc of VTKHDF file format and the VTK book in these examples.

The goal of this repository is to store python script to help user to use VTKHDF File Format. Reducing the number of dependency is better for taht.

So the stated goal in the description of the issue is very different from the one you're telling me now: the goal of this repository is giving users of VTKHDF examples of how to generate VTKHDF data without starting from a VTK dataset. This implies that each H5 dataset written is explained so users know how to derive a VTKHDF from their data, wherever it comes from. Which in turn implies that they have some familiarity with the VTK data model. I agree that for now this repository does not a great job at that. Neither the README indicates this goal, nor the top level vtkxml-to-vtkhdf directory which does literally the opposite.

yes, I agree with your statement, it definitely should be improved.

@lucas.givord let's rename and improve this issue to reflect this new direction, this is not what I initially understood