ops._arith_method_FRAME typo?

[jbrockmendel]

https://github.com/pandas-dev/pandas/blob/master/pandas/core/ops.py#L1030

                # we may need to manually
                # broadcast a 1 element array
                if yrav.shape != mask.shape:
                    yrav = np.empty(mask.shape, dtype=yrav.dtype)
                    yrav.fill(yrav.item())

Getting yrav.item() in the last line here after re-defining yrav in the previous line seems like it is likely getting the wrong thing. Is this intentional?

[jbrockmendel]

It looks like in the tests this branch is only hit in py3 and in one test:

s = Series([2, 3, 4, 5, 6, 7, 8, 9, datetime(2005, 1, 1)])
s[::2] = np.nan
d = DataFrame({'A': s})

with pytest.raises(ValueError):
    d.__and__(s, axis='columns')

... and the reason why a ValueError is raised ATM is because the call to yrav.item() raises ValueError: can only convert an array of size 1 to a Python scalar

So still not sure what's going on, but increasingly confident this isn't how it is supposed to work.