Github user cloud-fan commented on the issue: https://github.com/apache/spark/pull/19862 I'm not sure we should spend time to optimize non-whole-stage-codegen path, which is only used as a safe fallback if whole stage codegen is not applicable. I think the more important thing to do is to support whole stage codegen for more queries. cc @hvanhovell
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org