Breaking master twice in one day is pretty bad. The CI in PR was green though :)

The fix for different signatures is in https://github.com/rth/scikit-learn/pull/new/annotations-openml (trying to add type annotations in general). Looks like we are not running linters on the merged version of the code.

Or you can just add # type: ignore for now. We should put mypy in a job that doesn't block CI. Maybe GH actions.

The CI in PR was green though :)

I know lol. I think the PR got merged before the mypy check got introduced, so mypy wasn't run on that PR I think the openml PR was ready before we introduced mypy in the CI, so we couldn't catch that mypy would have failed on it

WDYT of the proposed fix here?

Or you can just add # type: ignore for now.

Oh yes that's better, did that

We should put mypy in a job that doesn't block CI. Maybe GH actions.

Yes good idea

Thanks @rth, linting passing so I'll merge since this is blocking other PRs

what a sunday lol

It is a lovely sunday xD

And now we have another issue in CircleCI:

! pdfTeX error (ext4): \pdfendlink ended up in different nesting level than \pd
fstartlink.

Anyone interested in debugging LaTeX? :)