SEZ9 commented on PR #11077: URL: https://github.com/apache/seatunnel/pull/11077#issuecomment-6961753067
Thanks for the update. Here is where the earlier points stand after the re-review of `352e5a27`. **Progress so far** - **Destination-key collisions:** writers are now grouped by connector class, destination identity and replica. - **Snapshot fan-out:** shared snapshots are now persisted once under a primary identifier instead of being fanned out to every alias. - **Schema divergence:** the `MultiTableSink` constructor now validates equal physical row layouts for equal destination keys. - **Docs and tests:** the documentation correction and the two requested tests are present. **Still open** 1. **Destination-key collisions:** the filesystem-option hash collision is still a concrete case. Distinct destination identities must never share a writer, so please make the key collision-proof or fail fast on a collision, and add a test for it. 2. **Shared-writer state on restore:** a shared writer's checkpoint must remain findable after the tables are reordered or an alias is removed. Please show how the primary identifier is resolved in those two cases and cover both with a restore test. 3. **Schema/config divergence:** the row-layout check covers schema. Is config divergence between aliases of the same destination also guarded, or is it intentionally left to the connector's identity? **Please confirm** - Does the documentation now cover both `getPhysicalDestinationIdentifier()` and the changed `restoreWriter` contract (merged state from all aliased identifiers passed to one call) for connector implementers? - Does every aliased identifier now get a `proxyContexts` entry, and is the `containsValue` scan gone? - Does an `IOException` from `createWriter`/`restoreWriter` now propagate as declared rather than being wrapped in a `RuntimeException`? - Were the missing param and return tags added to the `getDestinationKey` Javadoc? A passing rerun alone won't clear items 1 and 2; they need the source changes and tests above. <!-- streview-comment:2407 --> -- 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]
