jayzhan211 commented on code in PR #27529: URL: https://github.com/apache/datafusion/pull/27529#discussion_r4614647895
########## datafusion/sqllogictest/test_files/repartition.slt: ########## @@ -139,3 +139,74 @@ statement ok DROP TABLE t1; # End repartition on empty columns test + +# Start reorder aggregate keys with duplicate group column names test. +# See https://github.com/apache/datafusion/issues/27221 + +# Force partitioned hash joins so the join reorders the aggregate's keys +statement ok +set datafusion.optimizer.hash_join_single_partition_threshold = 0; + +statement ok +set datafusion.optimizer.hash_join_single_partition_threshold_rows = 0; + +statement ok +CREATE TABLE t1 (a INT) AS VALUES (1), (2), (3); + +statement ok +CREATE TABLE t2 (a INT) AS VALUES (10), (20), (30); + +statement ok +CREATE TABLE t3 (a INT) AS VALUES (100), (200), (300); + +statement ok +CREATE TABLE t4 (k1 INT, k2 INT, k3 INT, v VARCHAR) AS VALUES + (1, 20, 300, 'x'), + (2, 30, 100, 'y'), + (3, 10, 200, 'z'); + +# Two keys swapped Review Comment: Optional: an `EXPLAIN` here would pin the restoring projection (`a@1 as a1, a@0 as a2`). Without it, these result-only tests would still pass if the join stopped being partitioned or the key reordering was skipped, and the fix would no longer be exercised. Expected output generated on this branch with `--complete`: ```suggestion query TT EXPLAIN SELECT g.a1, g.a2, t4.v FROM (SELECT t1.a AS a1, t2.a AS a2 FROM t1 CROSS JOIN t2 GROUP BY t1.a, t2.a) g JOIN t4 ON g.a2 = t4.k2 AND g.a1 = t4.k1; ---- logical_plan 01)Projection: g.a1, g.a2, t4.v 02)--Inner Join: g.a2 = t4.k2, g.a1 = t4.k1 03)----SubqueryAlias: g 04)------Projection: t1.a AS a1, t2.a AS a2 05)--------Aggregate: groupBy=[[t1.a, t2.a]], aggr=[[]] 06)----------Cross Join: 07)------------TableScan: t1 projection=[a] 08)------------TableScan: t2 projection=[a] 09)----TableScan: t4 projection=[k1, k2, v] physical_plan 01)HashJoinExec: mode=Partitioned, join_type=Inner, on=[(a2@1, k2@1), (a1@0, k1@0)], projection=[a1@0, a2@1, v@4] 02)--ProjectionExec: expr=[a@1 as a1, a@0 as a2] 03)----AggregateExec: mode=FinalPartitioned, gby=[a@0 as a, a@1 as a], aggr=[] 04)------RepartitionExec: partitioning=Hash([a@0, a@1], 4), input_partitions=1 05)--------AggregateExec: mode=Partial, gby=[a@1 as a, a@0 as a], aggr=[] 06)----------CrossJoinExec 07)------------DataSourceExec: partitions=1, partition_sizes=[1] 08)------------DataSourceExec: partitions=1, partition_sizes=[1] 09)--RepartitionExec: partitioning=Hash([k2@1, k1@0], 4), input_partitions=1 10)----DataSourceExec: partitions=1, partition_sizes=[1] # Two keys swapped ``` -- 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]
