[
https://issues.apache.org/jira/browse/SPARK-55299?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18081578#comment-18081578
]
Anupam Yadav commented on SPARK-55299:
--------------------------------------
[~gaogaotiantian] I would like to help with this, if you think that's okay?
wanted to clarify a few questions about the expected behavior:
you mentioned Pandas 3 has this change, are you talking about the default
resolution change from `ns` to `us` for timedelta?
Also, should the fix propagate the unit from source timestamps? so for example
if both timestamps are `ns` then the resulting timedelta should also be `ns`?
Happy to take direction on the implementation. Thanks!
> Guess the correct unit for timedelta if it's calculated
> -------------------------------------------------------
>
> Key: SPARK-55299
> URL: https://issues.apache.org/jira/browse/SPARK-55299
> Project: Spark
> Issue Type: Sub-task
> Components: PySpark
> Affects Versions: 4.2.0
> Reporter: Tian Gao
> Priority: Major
>
> When timedelta is calculated from timestamps, the unit should follow. Pandas
> 3 has this change I think. We need to take care of it.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
