adriangb commented on code in PR #18451:
URL: https://github.com/apache/datafusion/pull/18451#discussion_r2535638143
##########
datafusion/core/tests/physical_optimizer/filter_pushdown/mod.rs:
##########
@@ -1667,8 +1667,8 @@ async fn test_nested_hashjoin_dynamic_filter_pushdown() {
- HashJoinExec: mode=Partitioned, join_type=Inner, on=[(a@0, b@0)]
- DataSourceExec: file_groups={1 group: [[test.parquet]]},
projection=[a, x], file_type=test, pushdown_supported=true
- HashJoinExec: mode=Partitioned, join_type=Inner, on=[(c@1, d@0)]
- - DataSourceExec: file_groups={1 group: [[test.parquet]]},
projection=[b, c, y], file_type=test, pushdown_supported=true,
predicate=DynamicFilter [ b@0 >= aa AND b@0 <= ab ]
- - DataSourceExec: file_groups={1 group: [[test.parquet]]},
projection=[d, z], file_type=test, pushdown_supported=true,
predicate=DynamicFilter [ d@0 >= ca AND d@0 <= cb ]
+ - DataSourceExec: file_groups={1 group: [[test.parquet]]},
projection=[b, c, y], file_type=test, pushdown_supported=true,
predicate=DynamicFilter [ CASE hash_repartition % 1 WHEN 0 THEN b@0 >= aa AND
b@0 <= ab ELSE false END ]
Review Comment:
I don't think nulls end up factoring in: nulls never get included in an
inner join (the only case for which this optimizations is active now). If we
want to support e.g. left joins I agree we'll have to modify the generated
filter for those cases to be what you are suggesting.
--
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]
