shouweikun edited a comment on pull request #14727: URL: https://github.com/apache/flink/pull/14727#issuecomment-770787085
> Hi @shouweikun , I have went through the pull request. However, supporting sink parallelism for Hive and Filesytem is not just changing parallelism of the writer DataStream. We should first support `ParallelismProvider` for `DataStreamSinkProvider` first. Because if the sink parallelism is different than the upstream operator, we should implicitly add a keyby shuffle if there is changelog in the stream, otherwise the changelog will be out of order. See > > https://github.com/apache/flink/blob/95257a255f0da0a95b31647c6d057914d5748871/flink-table/flink-table-planner-blink/src/main/java/org/apache/flink/table/planner/plan/nodes/exec/common/CommonExecSink.java#L116 Coincidently, the code quoted above was written by me (LOL). Actually, the aim of this exception is trying to prevent user from making `DataStreamSinkProvider` inherit `ParallelismProvider`. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
