Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.40%. Comparing base (c96e1dd) to head (89e1d6e).

@@           Coverage Diff           @@
##             main    #7885   +/-   ##
=======================================
  Coverage   96.40%   96.40%           
=======================================
  Files         150      150           
  Lines       32675    32675           
  Branches     4115     4115           
=======================================
  Hits        31501    31501           
  Misses        570      570           
  Partials      604      604           

🚀 Deployed on https://68bf0a635b126b08e0b310e6--pyvista-dev.netlify.app

An attempt of using parallel jobs for testing was already proposed in another PR and led to the same conclusion, ie. that downloadable datasets must occur within fixtures to prevent such kind of issues
See #7624 (comment)

An attempt of using parallel jobs for testing was already proposed in another PR and led to the same conclusion, ie. that downloadable datasets must occur within fixtures to prevent such kind of issues
See #7624 (comment)

That's the second time I've neglected to use that PR as a basis. Seems like there are several changes that are needed to get pytest-xdist working and we might have to break this out into several PRs. I did notice that the main issue was on collection, but it's possible that it would crop up given enough runs.

Possible unstable test test_direction_objects with vtk==9.3.1, see job=49561617651.

test_axes_actor_default_colors failed in Linux Unit Testing (3.11, 9.2.6, latest).

Ran cluster local testing on docker and discovered that <9.4 segfaults on render, but I've not seen any failures on vtk>=9.4 despite repeatably running tests with:

xvfb-run pytest tests/plotting -n24

This matches the failures we've had here so far.

I'm not seeing any difference in the render window used as it's identical across all versions of VTK on our CI, so there might be another change under the hood that's causing this.

        GPU Vendor : Mesa
      GPU Renderer : llvmpipe (LLVM 15.0.7, 256 bits)
       GPU Version : 4.5 (Core Profile) Mesa 23.2.1-1ubuntu3.1~22.04.3
     Render Window : vtkRenderWindow

On Linux, we might be able to reliably run parallel testing for VTK >= 9.4 on Linux, and parallel core testing for all.

Windows appears rock solid for the past several runs.