Github user heary-cao commented on the issue: https://github.com/apache/spark/pull/19143 @srowen Actually, I think this one makes sense. Because the two open test cases are tested, we connect the Cartesian product and the broadcast nested loop join. When we understand the two connections, from the side, we can open these two test cases to find out. Per Row (NS), for example, is much more delayed than our previous hash join. Second, in our JoinBenchmark test, the two test cases are missing. I wonder if you agree with this?
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org