Github user joseph-torres commented on a diff in the pull request: https://github.com/apache/spark/pull/18790#discussion_r131761258 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala --- @@ -872,6 +886,25 @@ object PushDownPredicate extends Rule[LogicalPlan] with PredicateHelper { pushDownPredicate(filter, u.child) { predicate => u.withNewChildren(Seq(Filter(predicate, u.child))) } + + case filter @ Filter(condition, watermark: EventTimeWatermark) => --- End diff -- For filter, the logic has a subtle additional condition. We don't want to push down filters on the watermark attribute, because: * they'll be at least as expensive to evaluate as the watermark operator * partition pruning shouldn't apply since there won't be useful partitions on an event time For project, I don't see a rule for UnaryNode anywhere. I might have missed it. For limit, I actually removed the rule for EventTimeWatermark that I originally added, since it does drop rows in some situations. So I don't think that making EventTimeWatermark subclass UnaryNode would avoid any of the code in this PR. I agree it seems appropriate, but it also seems orthogonal.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org