rgbuilds commented on PR #26441: URL: https://github.com/apache/datafusion/pull/26441#issuecomment-6410122541
@andygrove Thanks for the detailed review and prototype. I’ve rebased onto current `main` and addressed the blocking issue. The value-grouped path now: - builds the pruning predicate once; - forms groups and split points from all listed files before pruning; - prunes independently within each group; and - retains empty groups so the declared range partitioning remains stable. I added the suggested A–D regression coverage for pruning B from both the dimension and fact sides. Both queries return A, C, and D, preserve `[A], [], [C, D]`, and avoid an additional `RepartitionExec`. I also removed the debug output, fixed the function spacing, resolved the `repartition_subset_satisfaction.slt` expectations after the rebase, and disabled statistics collection around the affected ClickBench `EXPLAIN` cases to retain their scan/aggregate plan-shape coverage. The focused tests, full SQL logic suite, formatting, and strict all-target/all-feature Clippy check pass. I left the two explicitly non-blocking follow-ups—planning-time pruning metrics and batching files into one `PrunableStatistics` call—for separate work, since they require broader metrics/shared-pruning changes. -- 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]
