cloud-fan commented on code in PR #55891:
URL: https://github.com/apache/spark/pull/55891#discussion_r3258764035
##########
core/src/main/java/org/apache/spark/util/collection/unsafe/sort/UnsafeExternalSorter.java:
##########
@@ -647,6 +678,11 @@ int getSpillMergeRounds() {
return boundedMerger == null ? 0 :
boundedMerger.getIntermediateRoundsCompleted();
}
+ @VisibleForTesting
+ int getSpillWritersCount() {
Review Comment:
`testBoundedMergeSnapshotIsolatedFromConcurrentSpill` uses this accessor in
one place -- comparing live-list size before/after the external spill. The
suite fixture already tracks every created spill file in `spillFilesCreated`
(lines 59/111), and `spillFilesCreated.size()` would give the same +1 evidence
that the routing fired. The signals differ in principle (current live-list size
vs. cumulative files-created count), but they agree in this test because
nothing removes from `spillWriters` here. Worth dropping this
`@VisibleForTesting` accessor if you don't see a near-term need for the
distinct signal?
--
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]
