voonhous commented on PR #19456: URL: https://github.com/apache/hudi/pull/19456#issuecomment-5158380218
Re-reviewed at `40261aca65ce`. All 11 points addressed; I checked each against the code rather than the replies. The three new tests are not decorative -- I reintroduced the exact regression each claims to catch: - dedup key narrowed back to base-index-only -> **only** `testRemapPredicateKeepsBothProjectionsOfOneBaseColumn` fails (the `c0#f` domain is silently dropped). `testRemapPredicateKeepsOneDomainPerPhysicalColumn` stays green, so the two pin distinct properties rather than overlapping. - predicate remap disabled -> `testPredicateOnColumnWithStaleMetastoreOrdinal` fails `0 rows` vs `5 expected`, i.e. the real #19387 symptom, now caught end-to-end through a `QueryRunner`. - merge-path `enablePredicatePushDown` flipped to `true` at `HudiPageSourceProvider:254` -> `testPredicateIsNotPushedIntoTheBaseReadOfAMergedSplit` fails. All green at head (19 unit + the two e2e), `test-hudi-trino-plugin` passing. Nothing outstanding from me. -- 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]
