[SPARK-31620][SQL] Fix reference binding failure in case of an final agg contains subquery #720
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…agg contains subquery Instead of using `child.output` directly, we should use `inputAggBufferAttributes` from the current agg expression for `Final` and `PartialMerge` aggregates to bind references for their `mergeExpression`. When planning aggregates, the partial aggregate uses agg fucs' `inputAggBufferAttributes` as its output, see https://github.com/apache/spark/blob/v3.0.0-rc1/sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/AggUtils.scala#L105 For final `HashAggregateExec`, we need to bind the `DeclarativeAggregate.mergeExpressions` with the output of the partial aggregate operator, see https://github.com/apache/spark/blob/v3.0.0-rc1/sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/HashAggregateExec.scala#L348 This is usually fine. However, if we copy the agg func somehow after agg planning, like `PlanSubqueries`, the `DeclarativeAggregate` will be replaced by a new instance with new `inputAggBufferAttributes` and `mergeExpressions`. Then we can't bind the `mergeExpressions` with the output of the partial aggregate operator, as it uses the `inputAggBufferAttributes` of the original `DeclarativeAggregate` before copy. Note that, `ImperativeAggregate` doesn't have this problem, as we don't need to bind its `mergeExpressions`. It has a different mechanism to access buffer values, via `mutableAggBufferOffset` and `inputAggBufferOffset`. Yes, user hit error previously but run query successfully after this change. Added a regression test. Closes apache#28496 from Ngone51/spark-31620. Authored-by: yi.wu <[email protected]> Signed-off-by: Wenchen Fan <[email protected]>
rshkv
changed the title
robert3005
approved these changes
Dec 8, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Upstream
Taking fix for failures in which Spark will fail to bind a column if an aggregation contains a subquery.
What changes were proposed in this pull request?
Mostly a cherry-pick. I had to introduce the
BaseAggregateExectrait (it came from a different PR) the one big method it has is cherry-picked from upstream.How was this patch tested?
Upstream PR introduced test.