Github user hvanhovell commented on the issue: https://github.com/apache/spark/pull/20699 A very valid use cases for repartitioning is limiting concurrency. For example when you have a UDF that calls some service you don't want to DDOS that service because you make requests from too many partitions. In its current form this PR doesn't break this pattern, but I do wonder where this in its current form is useful.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org