fatmanverse commented on PR #11516: URL: https://github.com/apache/seatunnel/pull/11516#issuecomment-6285563419
Thanks @SEZ9 for the review and @DanielLeens for the summary and the approval. All four points are addressed in f4e693b: 1. **Hive multi-table behavior change (F1):** the PR description now has a section on it: Hive multi-table Parquet rows on Zeta / Flink carry the split's table path instead of `null`, like the other formats. `testParquetReadPreservesTableId` reads with a table path, as `MultipleTableHiveSourceReader` does, and asserts it on every row. 2. **Spark guard (F2):** I dropped the guard in `InternalRowConverter.parcel` and its test, as you both suggested. With the `ParquetReadStrategy` fix no in-tree caller hands a null table id to this path: `BaseFileSourceReader` passes `""`, and the multi-table readers pass the table path of the split. 3. **Test coverage (F3):** `testParquetRead1` now asserts the `""` id that `BaseFileSourceReader` passes, and `testParquetReadColumnNameNotCompatibleWithAvro` (which goes through the native parquet fallback reader) asserts the table id too. 4. **Guard comment (F4):** no longer applies, since the guard is removed. Verified on JDK 17: `ParquetReadStrategyTest` 17 tests and `MultiTableManagerTest` 7 tests pass. Could you take another look when you have time? Thanks! -- 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]
