zhuxiangyi commented on PR #10144: URL: https://github.com/apache/paimon/pull/10144#issuecomment-5807250596
Thanks for the review, both points are addressed in `81ac2b5`. **[P1] Java 8.** Reproduced with javac 1.8 and fixed by using `ClassLoader.getSystemClassLoader().getParent()`. Removing the guard in `NestedFieldReferences` still makes the test fail with `NoClassDefFoundError`, so the hidden-class check is intact. **Pruning assertion.** `NestedFieldFilterPushDownITCase.testNestedPredicatePrunesReads` writes two files whose `s.a` ranges do not overlap and counts what comes back through `newRead()` without `executeFilter()`, so nothing filters row by row: | predicate | rows read | | --- | --- | | none | 6 | | `s.a = 2` | 3 | | `s.a = 9999` | 0 | | `s.a > 0` | 6 | `s.a = 2` returning 3 rather than 1 is the point: the second file's row group is skipped whole and the first file's three rows all come back, which is row-group pruning and not row-level filtering. The 0 and 6 cases pin both ends. The MiniCluster starts here, so the integration tests do run: 128 tests pass on Flink 1.20 and again with `-Pflink2` on Flink 2.2. -- 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]
