kosiew commented on code in PR #24192:
URL: https://github.com/apache/datafusion/pull/24192#discussion_r3800963206
##########
datafusion/physical-plan/src/joins/sort_merge_join/tests.rs:
##########
@@ -5060,6 +5136,113 @@ async fn bitwise_multi_spill_inner_key_group() ->
Result<()> {
// 100-byte pool: every buffered slice fails its reservation, so each
// inner batch of the key group spills separately.
+ let runtime = RuntimeEnvBuilder::new()
+ .with_memory_limit(100, 1.0)
+ .with_disk_manager_builder(
+
DiskManagerBuilder::default().with_mode(DiskManagerMode::OsTmpDirectory),
+ )
+ .build_arc()?;
+
+ for (join_type, expected_rows) in [(LeftSemi, 1), (LeftAnti, 0),
(LeftMark, 1)] {
Review Comment:
Nice coverage here. It might also be worth adding a mirrored multi-spill
fixture for `RightSemi`, `RightAnti`, and `RightMark`. Those variants buffer
`left` while streaming `right`, so the two-batch equal-key group would need to
be on `left`. Adding them directly to this fixture would not exercise
multi-spill retention since its `left` input only has one row. The new no-spill
test already covers the error path for these variants, so this is just a
non-blocking suggestion.
--
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]
