SEZ9 commented on PR #11077: URL: https://github.com/apache/seatunnel/pull/11077#issuecomment-6137144355
Thanks for the rebase onto `dev`, @hesam-oxe — good to hear `c213b361` is mergeable again and that the `incompatible-changes.md` conflict in `docs/en` and `docs/zh` was resolved by keeping all three Connector Changes entries. Please share the outcome of the local `multitablesink` unit-test run once it finishes; full E2E can stay with CI as you said. Since the shared-writer scope is unchanged since the last pass, could you confirm per earlier point whether it is addressed in `c213b361` or still open? - **PR11077-F1 (HIGH)** — `DestinationKey` collisions routing one table's rows through another destination's writer. Is there now a guard/validation so two genuinely different destinations can't collapse into one shared writer? - **PR11077-F2 (HIGH)** — snapshot fan-out of shared-writer state to every aliased identifier plus the restore-time union duplicating that state N times. How is state deduplicated on restore now? - **PR11077-F3** — the shared writer being built from an arbitrary alias's sink/`CatalogTable` while receiving rows from all aliases. Is schema/config divergence detected, or documented as a connector responsibility? - **PR11077-F4 / PR11077-F6 (Docs)** — documenting the new `getPhysicalDestinationIdentifier()` SPI method and the changed `restoreWriter` contract (merged state from all aliased identifiers delivered in one `SeaTunnelSink#restoreWriter` call). Where do connector implementers find that now? - **PR11077-F5** — `proxyContexts` only registering the first alias per destination via `containsValue`. Does every aliased identifier now get a context entry, and is the O(n²) lookup gone? - **PR11077-F7** — `IOException` from `createWriter`/`restoreWriter` wrapped in an unchecked `RuntimeException` inside `computeIfAbsent`. Has this been restructured so the declared checked exception still propagates? - **PR11077-F8 (LOW)** — missing parameter/return tags on the `getDestinationKey` Javadoc. A short checklist (fixed / open / disagree, with a one-line pointer each) is enough; I'll do the follow-up pass as soon as I have that plus the unit-test result. <!-- streview-comment:1709 --> -- 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]
