FIX: Avoid copying source script when plot_html_show_source_link is False in plot directive - #20814

This seems reasonable and correct to me.

Could you possibly add a few more asserts to https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/tests/test_sphinxext.py to make sure the source is and is not copied in the relevant cases?

The rst we are running against is https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/tests/tinypages/some_plots.rst

Thank you for working on this and welcome 😄 (sorry should have lead with that!).

@tacaswell Added a few asserts. Could you take another look?

This seems reasonable, however I have a concern that this is going to add time to the test process as tinypages was already the slowest test:

On a recent default-branch run:

https://github.com/matplotlib/matplotlib/runs/3301881275?check_suite_focus=true#step:12:149

29.61s call     lib/matplotlib/tests/test_sphinxext.py::test_tinypages

and on this PR https://github.com/matplotlib/matplotlib/pull/20814/checks?check_run_id=3301939622#step:12:149

36.72s call     lib/matplotlib/tests/test_sphinxext.py::test_tinypages

I assume we are hitting some caching in sphinx so the second run does not double the run time, but are still adding 5s (I'm a bit torn how much to trust these times though as CI runtime is notoriously variable).

I think it would be better to add another plot to the example file?

I think it would be better to add another plot to the example file?

plot_html_show_source_link is a global configuration, not a directive-level configuration, so I don't think we can do something like:

.. plot::
     :show_source_link: False
     
     # code

Can we create a directory containing an RST file, perform sphinx-build against this directory with plot_html_show_source_link =True, and then assert a source script has not been created?

plot_html_show_source_link is a global configuration, not a directive-level configuration,

🐑 Sorry for being wrong about that and pushing you to test something impossible!

@QuLogic Could you take another review?

Thanks @harupy! Congratulations on your first PR to Matplotlib 🎉 We hope to hear from you again.