Github user liancheng commented on a diff in the pull request:
https://github.com/apache/spark/pull/12040#discussion_r57913359
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/Window.scala ---
@@ -885,9 +886,15 @@ private[execution] object AggregateProcessor {
val evaluateExpressions =
mutable.Buffer.fill[Expression](ordinal)(NoOp)
val imperatives = mutable.Buffer.empty[ImperativeAggregate]
+ // SPARK-14244: `SizeBasedWindowFunction`s created on driver side and
serialized to executor
+ // side contains global window partition size attribute reference
(`SizeBasedWindowFunction.n`),
+ // and must be rewritten using the one instantiated on executor side.
Otherwise, attribute
+ // binding fails because of different expression ID.
+ val rewrittenFunctions = functions.map(rewriteWindowPartitionSize)
+
// Check if there are any SizeBasedWindowFunctions. If there are, we
add the partition size to
// the aggregation buffer. Note that the ordinal of the partition size
value will always be 0.
- val trackPartitionSize =
functions.exists(_.isInstanceOf[SizeBasedWindowFunction])
+ val trackPartitionSize =
rewrittenFunctions.exists(_.isInstanceOf[SizeBasedWindowFunction])
--- End diff --
Actually I have to resort to side effects as the extra `Expression`
constructor argument fails function lookup and the function registry always
falls back to Hive `GenericUDAFCumeDist`. That's the reason of the most recent
test failures.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
