[
https://issues.apache.org/jira/browse/SPARK-61867?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hyukjin Kwon updated SPARK-61867:
---------------------------------
Fix Version/s: connect-rust-4.4.0
> Give the protoc-generated modules of Spark Connect plugins the classes of the
> Rust protobuf runtime, so that Delta Lake's DML plans take the drop-in's
> relations
> ----------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: SPARK-61867
> URL: https://issues.apache.org/jira/browse/SPARK-61867
> Project: Spark
> Issue Type: Bug
> Components: Connect
> Affects Versions: connect-rust-4.4.0
> Reporter: Hyukjin Kwon
> Priority: Major
> Fix For: connect-rust-4.4.0
>
>
> Since SPARK-61786, pyspark.sql.connect.proto exports the messages of the
> drop-in's Rust protobuf runtime. A Spark Connect plugin's Python client, such
> as delta-spark 4.0.0's delta.connect, has protoc-generated modules that
> import Spark Connect's protos and build their messages with protobuf's own
> classes. Their fields of Spark Connect's types then refuse the drop-in's
> messages: DeltaTable.delete, update and merge raise TypeError at
> relation.delete_from_table.target.CopyFrom(self._target.plan(session)), where
> the official client works.
> Also, a plugin's relation or command that holds df._plan.plan(session), as
> Delta's merge does with its source, holds the drop-in's placeholder of the
> DataFrame's relation inside the extension's Any, which the core did not
> replace.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]