TST: div/truediv with PyArrow duration types

[phofl]

Looks like they implemented __truediv__ for more types.

https://github.com/pandas-dev/pandas/actions/runs/5707046701/job/15463354529

[mroeschke]

#54327 doesn't not effectively compare the result to the equivalent numpy operation because pd.Timedelta / pd.NA raises a TypeError. Would be good to have dedicated tests for this

[prasad-yashdeep]

take

[AdventurousDataScientist]

Hi @prasad-yashdeep are you still working on this?

[ccccjone]

take

[droussea2001]

Hello @ccccjone are you still working on it ?

[ccccjone]

Hello @ccccjone are you still working on it ?

Hi @droussea2001, I've unassigned myself from this issue. Please feel free to take it on.

[droussea2001]

Ok thanks !

[droussea2001]

take

[droussea2001]

Hello,

I am a total beginner in open source and I am not sure to understand the issue. :-)

If my understanding is correct the issue appears from the following test functions:

- test_arith_series_with_scalar
- test_arith_frame_with_scalar
- test_arith_series_with_array

"#54327 doesn't not effectively compare the result to the equivalent numpy operation because pd.Timedelta / pd.NA raises a TypeError" because:
_get_arith_xfail_marker set a pytest xfail with a TypeError before the operation (further in floordiv_compat)

So the trick would be to:

• desactivate the xfail in _get_arith_xfail_marker for specific "COMPAT: Pyarrow supports div with duration #54327 div / truediv operations
• set a correct exception in the _get_expected_exception function called by check_opname
• consequently pytest.raises(exc) in _check_op would catch the correct exception

@mroeschke: is it correct ?

Thanks in advance,