JingsongLi commented on PR #9966: URL: https://github.com/apache/paimon/pull/9966#issuecomment-5832590940
Reviewed head `ac9d087` for production. This fixes a real end-to-end Data Evolution read loss: when a projection keeps only a blob/vector file covering part of a row-ID group, the reader otherwise derives its output range from that partial file and silently omits the remaining rows. The change retains the full-range normal anchor only when the kept dedicated files do not cover the requested row range. The real table read test checks all 10 rows, including five NULL-filled rows outside the blob subrange; the scan tests cover full and partial row-range pushdown. Local JDK 8 verification passed: `DataEvolutionFileStoreScanTest` 19/19 and `BlobUpdateTest` 4/4. `git diff --check` passed. Exact-head JDK 8/11, Spark, Flink, E2E, and licensing CI are green. I found no production blocker; keep this PR open for merge. -- 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]
