Describe the bug, including details regarding any error messages, version, and platform.
In [16]: pa.array([datetime(2020, 1, 1, tzinfo=timezone(timedelta(hours=-8)))], type=pa.timestamp('us', 'US/Pacific'))
Out[16]:
<pyarrow.lib.TimestampArray object at 0x7fb4b56b93c0>
[
2020-01-01 08:00:00.000000
]
In [17]: pa.array([datetime(2020, 1, 1, tzinfo=timezone(timedelta(hours=-8)))], type=pa.timestamp('us', 'UTC'))
Out[17]:
<pyarrow.lib.TimestampArray object at 0x7fb4b51426e0>
[
2020-01-01 08:00:00.000000
]
In [18]: pa.array([datetime(2020, 1, 1)], type=pa.timestamp('us', 'US/Pacific'))
Out[18]:
<pyarrow.lib.TimestampArray object at 0x7fb4b5142500>
[
2020-01-01 00:00:00.000000
]
Note how the result doesn't depend on the time zone I put in the type argument
Is this intentional? I'm not saying it's a bug or that it needs to change, it just came up in Polars: https://github.com/pola-rs/polars/issues/10662 - I just want to make sure that if there is any change, then we all go towards the same direction
Describe the bug, including details regarding any error messages, version, and platform.
Note how the result doesn't depend on the time zone I put in the
type
argumentIs this intentional? I'm not saying it's a bug or that it needs to change, it just came up in Polars: https://github.com/pola-rs/polars/issues/10662 - I just want to make sure that if there is any change, then we all go towards the same direction
versions:
Component(s)
Python