You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Jiangjie Qin (Jira)" <ji...@apache.org> on 2023/03/29 08:23:00 UTC
[jira] [Updated] (FLINK-31654) DataStreamUtils.reinterpretAsKeyedStream() should not override the user specified chaining strategy.
[ https://issues.apache.org/jira/browse/FLINK-31654?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jiangjie Qin updated FLINK-31654:
---------------------------------
Fix Version/s: 1.18.0
1.17.1
> DataStreamUtils.reinterpretAsKeyedStream() should not override the user specified chaining strategy.
> ----------------------------------------------------------------------------------------------------
>
> Key: FLINK-31654
> URL: https://issues.apache.org/jira/browse/FLINK-31654
> Project: Flink
> Issue Type: Bug
> Components: API / DataStream
> Affects Versions: 1.17.0, 1.14.6, 1.16.1, 1.15.4
> Reporter: Jiangjie Qin
> Priority: Major
> Fix For: 1.18.0, 1.17.1
>
>
> Currently {{DataStreamUtils.reinterpretAsKeyedStream()}} does not work well with batch jobs. There are two issues:
> # The input to the downstream operator will not be sorted. So users need to sort the records by themselves.
> # The result {{KeyedStream}} will still override the chaining strategy of the downstream operator to {{HEAD.}} This breaks the operator chain and defeats the purpose of reinterpreting the stream instead of calling {{keyBy.}}
> This ticket tends to address the second issue.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)