API: Handle pow & rpow special cases #30097
Conversation
| def test_rpow_special(value): | ||
| result = value ** pd.NA | ||
| assert result == 1 | ||
| if not isinstance(value, (np.float_, np.bool_, np.int_)): |
There was a problem hiding this comment.
Somebody (Cython?) is converting these to Python scalars.
There was a problem hiding this comment.
have you tried with/without numexpr?
There was a problem hiding this comment.
How can numexpr be involved in this?
There was a problem hiding this comment.
i dont know, but if we're brainstorming things that can do surprising conversions, it comes to minid
jreback
added
Missing-data
The NA scalar itself is currently documented in missing_data.rst. I would maybe add it there? |
|
Didn't see that section. Moved the note there. We will still need something in reference so that :class:`NA` points somewhere. |
|
This is blocking #29964, so it'd be nice to merge this in soon if possible. |
|
Addressed those, thanks for catching that. |
|
|
||
| @pytest.mark.parametrize( | ||
| "value", [1, 1.0, True, np.bool_(True), np.int_(1), np.float_(1)] | ||
| ) |
There was a problem hiding this comment.
Good catch.
In [4]: -1 ** np.nan
Out[4]: -1.0
In [5]: np.nan ** -0
Out[5]: 1.0Will match the NumPy behavior here.
| @@ -69,6 +72,49 @@ def test_comparison_ops(): | |||
| assert (other <= NA) is NA | |||
|
|
|||
|
|
|||
| @pytest.mark.parametrize( | |||
There was a problem hiding this comment.
There are fixtures defined for zero / one values in pandas/tests/indexing/conftest.py - can these be combined with that?
There was a problem hiding this comment.
That seems to include arrays. We just want a scalar here I think.
It's part of some IEEE spec, but I haven't tracked down the source. Both Python and NumPy use this behavior for NaN. For NA, I think it aligns with the principles behind Kleene logic. |
|
thanks @TomAugspurger |
Closes #29997
@jorisvandenbossche do you have thoughts on where to document this behavior? I added a bit to
reference/arrays.rstfor now, but perhaps we'll eventually have auser_guide/scalar_na.rst?