TST/CI: xfail test_round_sanity for 32 bit #47803
Conversation
mroeschke
added
CI
| @@ -297,6 +298,7 @@ def test_round_implementation_bounds(self): | |||
| with pytest.raises(OverflowError, match=msg): | |||
| Timestamp.max.ceil("s") | |||
|
|
|||
| @pytest.mark.xfail(not IS64, reason="Failing on 32 bit build", strict=False) | |||
There was a problem hiding this comment.
instead what if we just adjust the bounds by 1e9?
There was a problem hiding this comment.
So just to confirm
@given(val=st.integers(iNaT + 1 + 1e9, lib.i8max - 1e9))
There was a problem hiding this comment.
i guess technically should use 10**9 but yah
There was a problem hiding this comment.
I had to cap the max and min values to np.iinfo(np.in32) for the 32 bit build instead
There was a problem hiding this comment.
that seems like something really is wrong on the 32bit build, so maybe an xfail is appropriate after all?
There was a problem hiding this comment.
Okay. I will change it back to xfail(strict=False) because I think the hypothesis still doesn't fail this consistently with certain params.
mroeschke
force-pushed
the
flaky/32/rounding
branch
from
b57d287 to
01d72bf
Compare
|
Going to merge this to get the CI back to green. Will open an issue to investigate this issue. |
This comment was marked as resolved.
This comment was marked as resolved.
|
xref #47823 |
xfailing just to get the CI to green. Might be addressable in the future cc @jbrockmendel