a fair amount of the xfails are already like this

sure i think we should be consistent and change then all

@jreback Not certain, but I think you may be referring to using the pytest.mark.xfail decorator. The decorator has limitations, in the case shown here it can't be used with the fixture transformation_func.

A search for "applymarker" turns up empty across all our tests.

not at all search for

mark = pytest.mark.xfail(
                reason="BUG: it must be pd.Series([1, 1, 3, 4], dtype=np.int16"
            )
request.node.add_marker(mark)

for example

~/pandas$ grep -r add_marker pandas | wc
      49      99    4128

Thanks, request.applymarker is the same as request.node.add_marker, I'll stick to add_marker for consistency.

yeah i think @jbrockmendel started putting these in recently. we should change all occurernces, update the docs and put in a pre-commit hook (once all fixed).