geoffreyclaude opened a new issue, #26369: URL: https://github.com/apache/datafusion/issues/26369
The fully ordered multi-column aggregation path added in #25150 avoids hashing adjacent equal keys. Temporal, decimal, and fixed-size binary keys still fall back to hash-based key storage, even when all grouping expressions are ordered. Extend this path to the remaining scalar types whose adjacent Arrow equality matches the existing grouping builders: - Timestamps (all units and timezone metadata), dates, valid time units, durations, and intervals. - Decimal32/64/128/256, preserving precision and scale. - FixedSizeBinary at every valid width. Keep single-column specializations and the current handling of floating-point, dictionary, and nested keys. Add SQL benchmarks that run before and after the change, plus correctness coverage for nulls, batch boundaries, metadata, and partial/final aggregation. -- 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]
