I think we could simplify this test by running it inline, something like:

import warnings

import pytest

def test_raise_type_error_on_non_string_warning():
    with pytest.raises(TypeError, match="Warning message must be str"):
        with pytest.warns(UserWarning):
            warnings.warn(1)

    # Check that we get the same behavior with the stdlib, at least if filtering
    # (see https://github.com/python/cpython/issues/103577 for details)
    with pytest.raises(TypeError):
        with warnings.catch_warnings():
            warnings.simplefilter("always")
            warnings.warn(1)

Agreed, also:

• Please put this test in test_recwarn.py instead of creating a new file.
• Add a -> None type annotation.
• Add a docstring like Check pytest.warns validates warning messages are strings (#10865).

thanks, updated

However I'm not sure this should be a bugfix, I would argue this is more of an improvement.

thanks, updated

thanks, updated

Agreed, also:

• Please put this test in test_recwarn.py instead of creating a new file.
• Add a -> None type annotation.
• Add a docstring like Check pytest.warns validates warning messages are strings (#10865).