fix: #1304 add PD_LTE_23 #1306
fix: #1304 add PD_LTE_23 #1306
Conversation
| @@ -140,12 +143,24 @@ def test_truediv_pd_series() -> None: | |||
| check(assert_type(left.rdiv(c), pd.Series), pd.Series) | |||
|
|
|||
|
|
|||
| def test_path_div() -> None: | |||
| # GH 682 | |||
| folder = Path.cwd() | |||
There was a problem hiding this comment.
Same about the use of ensure_clean, the rest is good.
cmp0xff
force-pushed
the
hotfix/cmp0xff/1304
branch
2 times, most recently
from
948e493 to
935b475
Compare
|
/nightly |
|
/pandas_nightly |
|
Looks like it is failing for nightly run with pandas 3.0 for the test_truediv_path, see https://github.com/pandas-dev/pandas-stubs/actions/runs/16727359383 |
|
I see the issue, in pandas 3.0 the default type is StringDtype and not object which does not support the div operator. If you do |
I tested locally like that, and ALL of those tests are failing with the nightly. Thanks @loicdiridollou for remembering the right "/" command! |
Yes, all of those tests need to be wrapped this way until the bug is fixed in pandas. That's why the comment as a ref to the pandas Github issue. |
|
|
||
| check(assert_type(left.truediv(p), pd.Series), pd.Series, Path) | ||
| check(assert_type(left.div(p), pd.Series), pd.Series, Path) | ||
| if PD_LTE_23 or not WINDOWS: |
There was a problem hiding this comment.
The or not WINDOWS should not be necessary.
cmp0xff
force-pushed
the
hotfix/cmp0xff/1304
branch
from
e9877aa to
ce7498a
Compare
|
/pandas_nightly |
|
Hi @Dr-Irv , it seeems that I cannot trigger the nightly test. |
|
You have triggered them @cmp0xff, see https://github.com/pandas-dev/pandas-stubs/pull/1306/checks?check_run_id=47350577240 |
There was a problem hiding this comment.
One minor thing to confirm but all CI pass so good on my end, thanks @cmp0xff !
| check(assert_type(left.rtruediv(p), pd.Series), pd.Series, Path) | ||
| check(assert_type(left.rdiv(p), pd.Series), pd.Series, Path) | ||
| # mypy thinks it's `Path`, in contrast to Series.__rtruediv__(self, other: Path) -> Series: ... | ||
| check(assert_type(tmp_path / fpaths, pd.Series), pd.Series, Path) # type: ignore[assert-type] |
There was a problem hiding this comment.
@Dr-Irv will confirm but I think this is an issue we have had in the past with mypy getting confused
|
|
||
| check(assert_type(left.rtruediv(p), pd.Series), pd.Series, Path) | ||
| check(assert_type(left.rdiv(p), pd.Series), pd.Series, Path) | ||
| # mypy thinks it's `Path`, in contrast to Series.__rtruediv__(self, other: Path) -> Series: ... |
There was a problem hiding this comment.
I tried to come up with a simple example to illustrate the issue and couldn't.
There was a problem hiding this comment.
Issue comes from how mypy reads the Path.__truediv__ typing and it prioritizes it over Series.rtruediv, not expert enough to know why
There was a problem hiding this comment.
Yes, I don't know why it did that. Tried some simple experiments and couldn't replicate it.
assert_type()to assert the type of any return value