Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
fix: Use the right row container to extract rows in grouping set (fac…
…ebookincubator#11899) Summary: Pull Request resolved: facebookincubator#11899 Currently we always extract rows from grouping set through the default row container in the grouping set. It can cause potential performance and even data correctness problem in spill path as the row extract depends on the column stats recorded in the row container to decide whether apply fast no null path or slow path with nulls. If we use default row container to extract rows from the row container used by spill, the column stats is not accurate. If spill merge row container has null, then we get random non-null value for non column. This is correctness problem. If spill merge row container has no null, then we get slow path with null handling This PR fixes by always use the right container to extract rows Reviewed By: Yuhta Differential Revision: D67352373 fbshipit-source-id: 9d48857c4b3ca270bfc6777e0c3e4fbc03ee1e03
- Loading branch information