SEZ9 commented on PR #11932: URL: https://github.com/apache/seatunnel/pull/11932#issuecomment-5629393016
Quick status update on the open threads against `b6c96b97356d`: **`Collector.restoreSchema` on Flink/Spark** — Agreed on treating this as a known Zeta-only limitation for now rather than a blocker, consistent with the `[Fix][Zeta]` scope. I'll file the follow-up issue for extending `restoreSchema` to the Flink/Spark translation-layer collectors and link it here. One small ask for this PR: a short Javadoc note on the default method saying it is currently only honoured by Zeta, so the connector-level restore in `IncrementalSourceReader` isn't read as working on every engine. **Docs / upgrade note** — Still on me. I'll verify the "Checkpoint restore compatibility" section against the current head (schema persisted in `JdbcSinkState`, and the non-exactly-once `JdbcSinkWriter` now emitting writer state) and resolve that thread once confirmed. **Re-check of the earlier `IncrementalSourceReader` / `JdbcSink` points** — Also on me. I'll go through the remaining items (empty `checkpointTables` guard, consistency of the deserializer- and collector-side restore gating, `TableSchema` selection in `restoreWriter`, serializability of the persisted schema, primary-key handling on a restored schema, and null-Xid states after an `exactly_once` flip) against the current head and post a one-line outcome per item. I'll ping here once those are closed out so we can do a final pass. Thanks for sticking with this one. <!-- streview-comment:972 --> -- 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]
