[
https://issues.apache.org/jira/browse/BEAM-8894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16993581#comment-16993581
]
Etienne Chauchot commented on BEAM-8894:
----------------------------------------
[discussion|[https://lists.apache.org/thread.html/bf340f975b78a5e4237cd2678bc9a09b0a47bb8adae2b0f7034c6749%40%3Cdev.beam.apache.org%3E]]
on heterogeneous coders in the output PCollection of flatten:
> Fix multiple coder bug in new spark runner
> ------------------------------------------
>
> Key: BEAM-8894
> URL: https://issues.apache.org/jira/browse/BEAM-8894
> Project: Beam
> Issue Type: Bug
> Components: runner-spark
> Reporter: Etienne Chauchot
> Assignee: Etienne Chauchot
> Priority: Major
> Labels: structured-streaming
> Time Spent: 10m
> Remaining Estimate: 0h
>
> The test below does not pass. I fails with an EOF while calling
> NullableCoder(BigEndianCoder.decode)
> 'org.apache.beam.sdk.transforms.FlattenTest.testFlattenMultipleCoders'
> In the "old" spark runner, this test passes because it never call the
> NullableCoder because there is no serialization done.
> There may be a problem in the NullableCoder itself
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
