Github user cloud-fan commented on a diff in the pull request:
https://github.com/apache/spark/pull/12719#discussion_r62454855
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
---
@@ -617,6 +618,46 @@ object NullPropagation extends Rule[LogicalPlan] {
}
/**
+ * Propagate foldable expressions:
+ * Replace all attributes with aliases of the original foldable
expressions except Union queries.
+ * Aliases and ordinal expressions are the main target to be transformed
after propagation. Other
+ * optimizations will take advantage of the propagated foldable
expressions.
+ *
+ * {{{
+ * SELECT 1.0 x, 'abc' y, Now() z ORDER BY x, y, 3
+ * ==> SELECT 1.0 x, 'abc' y, Now() z ORDER BY 1.0, 'abc', Now()
+ * }}}
+ */
+object FoldablePropagation extends Rule[LogicalPlan] {
+ def apply(plan: LogicalPlan): LogicalPlan = {
+ val isFoldableStatement = plan.find {
+ case _: Union => true
+ case _: Command => true
+ case _ => false
+ }.isEmpty
+
+ val foldableExprSet = ExpressionSet(plan.flatMap {
+ case Project(projectList, _) => projectList.collect {
+ case a: Alias if a.resolved && a.child.foldable => a
+ }
+ case _ => Nil
+ })
+
+ if (!isFoldableStatement || foldableExprSet.isEmpty) {
--- End diff --
So we won't do this optimization if there exists a un-foldable plan(`Union`
and `Command`) in the plan tree? I think we can make it more fine-grained, e.g.
we can still apply this optimization for the children of `Union`.
A possible approach is: find one foldable project, collect foldable
expressions, transformUp the plan tree and replace attributes until we reach
`Union` or `Command`
---
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
