Ruth Comer

It does seem like there was some kind of change to actions storage space around the end of February, but this is getting way out of my wheelhouse https://github.com/actions/runner/issues/3184

Thanks for the report @Clarmy. This looks like it could be the same problem as #2260. Please could you try again with the newly released Cartopy 0.23?

> Do we know why we are getting duplicate points/bad geometries from our other code paths? This is likely OK to fix here, but it might be nice to at...

I have opened an issue at https://github.com/contourpy/contourpy/issues/377, and in the process came up with an example that I think is small enough to make into a test (though I see...

> test case is superbly done I just took your data and sliced it as much as I could while still reproducing the error! It's great to have the example...

By the way, you can also install pre-commit locally to help with linting each time you commit: https://scitools.org.uk/cartopy/docs/latest/contribute.html

Just to note that I tried `make_valid` on @Clarmy's original bad polygon and it took about 50x longer than the buffer approach. Also `make_valid` only came in at Shapely v1.8...

contour/contourf used to create one path per contour as you say but there was a change at Matplotlib v3.8 so they now create one path per level. https://matplotlib.org/stable/api/prev_api_changes/api_changes_3.8.0.html#contourset-is-now-a-single-collection Note that...

I think your understanding is correct > if we directly obtain the paths from ContourPy, does it imply that this part of the code needs to be refactored? This is...

I tried @Clarmy's original example with an old environment we have lying around (Cartopy 0.18, Matplotlib 3.3.4) and the error is the same, so I don't think the recent `ContourSet`...