ohad-definity commented on PR #59706: URL: https://github.com/apache/spark/pull/59706#issuecomment-6491505772
@sunchao friendly ping when you have a moment. Since your review: - P1 and P2 are fixed in abb931f4, with regression tests on your counterexamples. - @IgorBerman added adversarial tests for array operators wrapping the generator input (sort_array, array_distinct, set ops, filter/array_sort, reverse/slice), with near-duplicate data so wrong pruning changes the rows. - Merged with master; CI is green. The one open point is the V2 builder lifecycle. Instead of a javadoc note, when the generator-path rewrite is rejected we could build the plain scan with `rebuildScan` on a fresh builder (as scan merging does), and skip the generator path when `hasBlockingPushdown`. Each builder then sees a single `pruneColumns` + `build()`. Does that work for you (cc @peter-toth)? Happy to implement it. -- 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]
