Cache examples

[mkundu1]

No description provided.

[akaszynski]

If possible, try including the build times for sphinx-gallery when building. See the build output from
https://github.com/pyansys/pymapdl/runs/6292105801?check_suite_focus=true

For example:

computation time summary:
    - ../../examples/00-mapdl-examples/2d_plate_with_a_hole.py:                                         0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/2d_pressure_vessel.py:                                           0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/3d_notch.py:                                                     0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/3d_plate_thermal.py:                                             0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/contact_elements.py:                                             0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/geometry.py:                                                     0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/mapdl_3d_beam.py:                                                0.00 sec   0.0 MB
    - ../../examples/00-mapdl-examples/mapdl_beam.py:                                                   0.00 sec   0.0 MB

This way, you can tell which examples are being built from scratch and if your cache is working.

[dnwillia-work]

If possible, try including the build times for sphinx-gallery when building. See the build output from https://github.com/pyansys/pymapdl/runs/6292105801?check_suite_focus=true

Looks like it's coming now? From the PR CI run:

Though, strange it takes 0 s to build each, :-). Ha, perhaps that is correct though because Sphinx did not build them since they were already there from before?

[mkundu1]

If possible, try including the build times for sphinx-gallery when building. See the build output from https://github.com/pyansys/pymapdl/runs/6292105801?check_suite_focus=true

Looks like it's coming now? From the PR CI run:

Though, strange it takes 0 s to build each, :-). Ha, perhaps that is correct though because Sphinx did not build them since they were already there from before?

This shows that the caching mechanism is working. There are 3 CI runs for this PR: The first one didn't have any cache, so it ran all examples. The second one didn't change any example script, so it uses the cache and skips the gallery build. The third one changes the example scripts which invalidates the cache, so it again ran the examples fully (there is an unittest failure in the latest run which I'm looking into).

[dnwillia-work]

This shows that the caching mechanism is working. There are 3 CI runs for this PR: The first one didn't have any cache, so it ran all examples. The second one didn't change any example script, so it uses the cache and skips the gallery build. The third one changes the example scripts which invalidates the cache, so it again ran the examples fully (there is an unittest failure in the latest run which I'm looking into).

Yeah, perfect then, this is going to help. It appears our CI runners still have some trouble with the parametric workflow: