cloud-fan commented on a change in pull request #32786:
URL: https://github.com/apache/spark/pull/32786#discussion_r646693103
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/AggregatingAccumulator.scala
##########
@@ -205,6 +205,10 @@ class AggregatingAccumulator private(
buffer = other.buffer
joinedRow = other.joinedRow
}
+
+ private[execution] def addState(v: InternalRow): Unit = {
Review comment:
I see, so we need a per-task `AggregatingAccumulator` instance and
keeping calling `add` during the task lifetime.
The problem is that we need a temp `AggregatingAccumulator` instance to
delay the accumulator updates until the end of the task, and it's a bit hard to
keep this temp `AggregatingAccumulator` per task.
Let me think more about this and see if I can come up with a better idea.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
