andygrove commented on code in PR #4818:
URL: https://github.com/apache/datafusion-comet/pull/4818#discussion_r4629039882


##########
spark/src/test/scala/org/apache/comet/exec/CometAggregateSuite.scala:
##########
@@ -3844,6 +3844,44 @@ class CometAggregateSuite extends CometTestBase with 
AdaptiveSparkPlanHelper {
     }
   }
 
+  test("kurtosis merges an empty partial like Spark") {

Review Comment:
   Moved in 4d4e573af. `kurtosis_partials.sql` uses the 
`percentile_partials.sql` layout, one file per partition. It holds the 
empty-partial case, and the Scala test is gone.
   
   It also merges ordinary values from two files, with a group whose first 
partial is empty. I picked the values so that rewriting the `avg`, `m2` or `m4` 
arithmetic in `moments4_merge` changes a result bit-exactly, and Spark gives 
the same answer in either partial arrival order. I checked each rewrite, and 
restoring the empty-partial skip, and the fixture fails for all four. The `m3` 
term is not covered. With two partials it can't change the result, and with 
three it only does in some arrival orders.
   



-- 
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]

Reply via email to