comphead commented on code in PR #6712:
URL: https://github.com/apache/datafusion-comet/pull/6712#discussion_r4231746578
##########
spark/src/test/scala/org/apache/comet/CometCodegenSuite.scala:
##########
@@ -969,6 +1006,39 @@ class CometCodegenSuite
}
}
+ test("identical non-deterministic dispatched expressions keep their own
state") {
Review Comment:
With `round_nondeterministic_child.sql` in place, this test checks the same
fix a second time. The `javaId` pair reaches the same tag in
`emitJvmCodegenDispatch` as the fixture's `round`, and per the PR description
the `idPassthrough` pair already passes without the fix. Separate Scala UDF
kernels keeping their own counters across batches are already covered by
`Nondeterministic state persists across two ScalaUDFs in one task`. Could the
fixture take over the cross-batch part with a `Config` line setting
`spark.comet.batchSize=2`, as `if_expr.sql` does, and this test be dropped? I
have not run it, but from reading `CometExecIterator` that setting becomes the
native session's batch size.
--
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]
