NoahKusaba commented on PR #2638: URL: https://github.com/apache/datafusion-ballista/pull/2638#issuecomment-6881484073
Thanks! Marked the docs page and the crate docs experimental, and added the limitation, with `ballista.planner.adaptive.enabled = false` as the workaround. One correction: on the pinned commit a plain `UNION ALL` now fails, but your `… ORDER BY id` repro still returns duplicate rows. The `SortExec` on each branch reads its scan whole, so #2721 never sees it. That's an adaptive-planner bug for every table format, tracked in #2854; #2852 turns it into an error. The limitation covers both, and `tests/scan.rs` now runs a split union under both planners. I also bumped the pins: Ballista to the last commit before DataFusion 56, since iceberg-datafusion is still on 55, and iceberg-datafusion to its latest main. -- 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]
