Elliott Sales de Andrade
Elliott Sales de Andrade
I'm assuming that this will, in general, be handled by the upgrade to FreeType 2.13 and the introduction of libraqm, as well as various previous changes to ttconv, etc.
Did this not work?
I think this is fixed by #27349.
This would be interesting, but would need to pass tests and linting. The What's new note could also use an example image, as could this PR to help motivate review.
Instead of repeating the category every time, why not do something like `[{"category": "The category.", "strings": ["1", "2, "3", ...]}, ...]`
> ```python > vertices = ellipse.get_vertices() > transformed_vertices = ellipse.get_transform().transform(vertices) > data_verts = ax.transData.inverted().transform(transformed_vertices) > ``` `ax.transData` will change based on plotted data; your calculated `data_verts` is going to be...
Are you interested in writing a test? It looks like you have a test script in #28793, and it should be able to be turned into a [figure comparison](https://matplotlib.org/devdocs/devel/testing.html#compare-two-methods-of-creating-an-image) if...
I don't understand; the point of the test would be to confirm that the shadow is set the way you'd expect it to, and thus have equal figures.
Well, there was no test before, so no confirmation that it ever did.
Well for now, since geopandas is noarch, I just brute-force retried until I got a builder that wasn't s390x.