[
https://issues.apache.org/jira/browse/SPARK-43415?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17879998#comment-17879998
]
Herman van Hövell commented on SPARK-43415:
-------------------------------------------
Moved to the unified scala epic.
I think the following will work. In case of a KVDS with mapValues and a typed
aggregations:
* Project keys, and values into an intermediate dataframe.
** We need to check if empty groups are allowed?
** We need to be a bit careful with RGDS to KVDS conversions.
* Aggregate by the key, and bind the aggregators to produced value column. We
will need to extend the ColumnNodeToProtoConversion for this (to include a
column and not just the encoder).
> Impl mapValues for KVGDS#mapValues
> ----------------------------------
>
> Key: SPARK-43415
> URL: https://issues.apache.org/jira/browse/SPARK-43415
> Project: Spark
> Issue Type: Improvement
> Components: Connect
> Affects Versions: 3.5.0
> Reporter: Zhen Li
> Priority: Major
>
> Use an resolved func to pass the mapValues together with all aggExprs. Then
> on the server side unfold it to apply mapValues first before running
> aggregate.
> e.g.
> https://github.com/apache/spark/commit/a234a9b0851ebce87c0ef831b24866f94f0c0d36
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
