Github user jiangxb1987 commented on the issue: https://github.com/apache/spark/pull/22112 > IMO we should traverse the dependency graph and rely on how ShuffledRDD is configured A trivial point here - Since `ShuffleDependency` is also a DeveloperAPI, it's possible for users to write a customized RDD that behaves like `ShuffleRDD`, so we may want to depend on dependencies rather than RDDs.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org