ericm-db commented on code in PR #58476:
URL: https://github.com/apache/spark/pull/58476#discussion_r3984449127
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/state/StateStore.scala:
##########
@@ -681,6 +681,14 @@ trait StateStoreInstanceMetric {
*/
def combine(originalMetric: SQLMetric, value: Long): Long
+ def combine(originalValue: Long, value: Long): Long = {
Review Comment:
Existing StateStoreInstanceMetric implementations are required to implement
combine(SQLMetric, Long), but the accumulator now calls this new overload
directly and its default is always max. A custom state-store provider whose
existing combine method uses min, sum, or another policy will continue to
compile but silently get max semantics when duplicate task attempts are merged.
Could we preserve the existing combine contract here, or explicitly establish
one canonical primitive-value method with compatibility handling? A regression
test using a metric whose combine policy differs from max would cover this.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
