Thanks for the pull request! However, the existing "array-like" is sufficient, see https://numpy.org/doc/2.1/glossary.html#term-array_like. A pandas DataFrame is covered by this. It's not feasible to write out any array-compatible datastructure explicitly.

We would only state DataFrame explicitly if we make use of specific DataFrame features like the index, as long as we only use the column values it's a generic array-like to us.

@timhoffm I talked to @thedatacurious about this PR a bit and it stemmed from a confusion with y b/c stackplot supports both the packed stackplot(x, y=[y1, y2, y3]) and the unpacked form stackplot(x, y1, y2, y3) form.

While this PR may be doing a bit too much in its current form, I think it's fair to be a bit more explicit on allowed inputs.

Sure, feel free to reopen and rephrase.

That said, I think we should more clearly specify what array-like means. But that's a separate topic (possibly a glossary entry when we revive that, and/or an :mpl-type: link).

@timhoffm Sure, it looks like I don't have permissions to reopen this PR, should I raise a new one?

I have rephrased the edits:
main...thedatacurious:matplotlib:doc-branch)

Please fix the trailing whitespace as reported by flake8.