Github user gczsjdy commented on the issue: https://github.com/apache/spark/pull/19862 This is actually a small change, but it can provide not small optimization for users who don't use `WholeStageCodegen`, for example there're still some users who use under 2.0 version of Spark. Also, the behavior of codegen and non-codegen code paths are supposed to be the same, so in this way it is a 'bug'.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org