BenoitHanotte opened a new pull request #10854: [FLINK-15577][table-planner] Fix similar aggregations with different windows being considered the same URL: https://github.com/apache/flink/pull/10854 ## What is the purpose of the change The RelNode's digest is used by the Calcite HepPlanner to avoid adding duplicate vertices to the graph. If an equivalent vertex was already present in the graph, then that vertex is used in place of the newly generated one. This means that the digest needs to contain all the information necessary to identifying a vertex and distinguishing it from similar (but not equivalent) vertices. In the case of the `WindowAggregation` nodes, the window specs are currently not in the digest, meaning that **two aggregations with the same signatures and expressions but different windows are considered equivalent by the planner, which is not correct and will lead to an invalid Physical Plan**. This commit fixes this issue and adds a test ensuring that the window specs are in the digest, as well as similar aggregations on two different windows will not be considered equivalent. Only the old planner is subject to the issue, the Blink planner correctly uses the window specs in the nodes' digests, allowing Blink to correctly differentiate between the nodes. More info and an example of an invalid plan are avalaible at https://issues.apache.org/jira/browse/FLINK-15577 ## Brief change log Added window specs to the following RelNodes: - LogicalWindowAggregate - FlinkLogicalWindowAggregate - LogicalWindowTableAggregate - FlinkLogicalWindowTableAggregate Added unit tests to the legacy planner and to the blink planner to prevent future regressions. ## Verifying this change This change added tests and can be verified as follows: - Added unit tests for the old planner to ensure window specs are in digest and that similar aggregations with different windows are not considered equivalent in the physical plan. - Added unit tests for the blink planner to ensure no such regression can be introduced in the future. ## Does this pull request potentially affect one of the following parts: - Dependencies (does it add or upgrade a dependency): no - The public API, i.e., is any changed class annotated with `@Public(Evolving)`: no - The serializers: no - The runtime per-record code paths (performance sensitive): no - Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Yarn/Mesos, ZooKeeper: no - The S3 file system connector: no ## Documentation - Does this pull request introduce a new feature? no - If yes, how is the feature documented? not applicable
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
