JingsongLi commented on PR #11286: URL: https://github.com/apache/paimon/pull/11286#issuecomment-7040958177
Thanks for checking the older Spark versions. I reran this head on master. The 13 Core tests and the 7 existing Spark system-table tests passed. An additional Spark 3.5.9 SQL test with the legacy flag enabled returned a value 28,800,000 ms earlier than the stored epoch for both system columns. The same test passed with master's system-table implementation. The existing legacy conversion now affects two more columns. Spark's legacy writer also converts SQL timestamps on input, so a global read-side LTZ change needs a round-trip compatibility check for existing tables. A system-table-specific mapping or a separate epoch-valued column would let us address the metadata issue without changing that write/read contract. The StarRocks freshness reader still needs to handle both schema versions before its dependency upgrade. -- 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]
