[GitHub] [arrow-ballista] Dandandan commented on a diff in pull request #664: Upgrade to DataFusion 18.0.0-rc1

["Dandandan (via GitHub)" <gi...@apache.org>]

Dandandan commented on code in PR #664:
URL: https://github.com/apache/arrow-ballista/pull/664#discussion_r1103634441


##########
ballista/scheduler/src/planner.rs:
##########
@@ -375,19 +373,19 @@ mod test {
         /* Expected result:
 
         ShuffleWriterExec: Some(Hash([Column { name: "l_returnflag", index: 0 }], 2))
-          AggregateExec: mode=Partial, gby=[l_returnflag@1 as l_returnflag], aggr=[SUM(l_extendedprice Multiply Int64(1))]
-            CsvExec: source=Path(testdata/lineitem: [testdata/lineitem/partition0.tbl,testdata/lineitem/partition1.tbl]), has_header=false
+          AggregateExec: mode=Partial, gby=[l_returnflag@1 as l_returnflag], aggr=[SUM(lineitem.l_extendedprice * Int64(1))]
+            CsvExec: files={2 groups: [[ballista/scheduler/testdata/lineitem/partition1.tbl], [ballista/scheduler/testdata/lineitem/partition0.tbl]]}, has_header=false, limit=None, projection=[l_extendedprice, l_returnflag]
 
         ShuffleWriterExec: None
-          ProjectionExec: expr=[l_returnflag@0 as l_returnflag, SUM(lineitem.l_extendedprice Multiply Int64(1))@1 as sum_disc_price]
-            AggregateExec: mode=FinalPartitioned, gby=[l_returnflag@0 as l_returnflag], aggr=[SUM(l_extendedprice Multiply Int64(1))]
-              CoalesceBatchesExec: target_batch_size=4096
-                UnresolvedShuffleExec
+          SortExec: [l_returnflag@0 ASC NULLS LAST]
+            ProjectionExec: expr=[l_returnflag@0 as l_returnflag, SUM(lineitem.l_extendedprice * Int64(1))@1 as sum_disc_price]
+              AggregateExec: mode=FinalPartitioned, gby=[l_returnflag@0 as l_returnflag], aggr=[SUM(lineitem.l_extendedprice * Int64(1))]
+                CoalesceBatchesExec: target_batch_size=8192
+                  UnresolvedShuffleExec
 
         ShuffleWriterExec: None
-          SortExec: [l_returnflag@0 ASC]
-            CoalescePartitionsExec
-              UnresolvedShuffleExec
+          SortPreservingMergeExec: [l_returnflag@0 ASC NULLS LAST]
+            UnresolvedShuffleExec

Review Comment:
   This depends on how many partitions it sorts into.
   As `SortPreservingMergeExec` is itself single threaded, after a certain number of partitions it will be faster to use `SortExec`.



-- 
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: github-unsubscribe@arrow.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org