Not really sure about the mypy tests. I think you need to update thewidget.pyi-file, probably with the default values so that the signatures match.

Also, I think it would be good to add a "smoke test" for this. That is, call one of the selectors without anonselect callback. I'd create a new test that doesn't have an argument foronselect and then add a comment like:

# Smoke test for creating selector without callback

before that call.

Oh, and reading#21929 I am not sure that this closes the issue. It is clearly a good contribution taking care of parts of it though.

Thank you for your work on this@thiagoluisbecker .

Could you also add a test of creatingSelector object without passing the onselect (to actually exercise the default path!).

Hi! Apologies for the delay. I understand the suggestions and I will implement them. Thanks for the answers.

ping@thiagoluisbecker. Are you still planning to work on this?

Hi! No.

This seems straightforward, so I've rebased, fixed the comments, added a what's new note, and modified tests to drop theonselect argument everywhere it usednoop (but not where it usedMock withnoop).

Also note that there's alsoSpanSelector that has anonselect parameter, but it can't be made optional because thedirection parameter comes after it.

Optional: We could deprecate for kw-only use for onselect and following arguments, which would make potential later removal simpler. But that can also be done in a follow-up.