EnricoMi opened a new pull request, #40334: URL: https://github.com/apache/spark/pull/40334
### What changes were proposed in this pull request? A `DataSourceV2` reporting a `KeyGroupedPartitioning` through `SupportsReportPartitioning` does not have to implement `HasPartitionKey`, and is thus not limited to a single key per partition. ### Why are the changes needed? Before Spark 3.3, `DataSourceV2` implementations could report a `ClusteredDistribution`, which allowed Spark to exploit the existing partitioning and avoid an extra hash partitioning shuffle step. Transformations like `groupBy` or window functions (with the right group / partitioning keys) would then be executed on the partitioning provided by the data source. ### Does this PR introduce _any_ user-facing change? This improves performance as existing partitioning is reused. ### How was this patch tested? Existing tests have been fixed. They falsely reported partitions with multiple keys via `HasPartitionKey`. -- 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: reviews-unsubscr...@spark.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
