DanielLeens commented on issue #11923: URL: https://github.com/apache/seatunnel/issues/11923#issuecomment-5369664827
Thanks for splitting this out and for checking the current `dev` implementation so carefully. I re-checked the current code path before replying, and the drift you listed is real: the docs section still reads like a concrete `MultiTableSinkWriter` API, but current `dev` routes rows by `tableId`, hashes against `blockingQueues.size()`, hands work to queues via `offerQueueElement(...)`, and also has the unknown-table / failure-policy branch that the snippet does not show. The `System.nanoTime() % replicaNum` example is also not a safe illustrative fallback, because it can produce a negative value. Given that, I would take option 1 as the maintainer direction for this issue: 1. rewrite the writer / replica-selection section so it matches the real current writer shape and names closely enough that readers can map the doc back to source; 2. if the docs intentionally simplify something for explanation, mark it explicitly as simplified instead of presenting it as a copy of the current API; 3. keep the narrower `Math.abs(Integer.MIN_VALUE)` correctness fix in #11721, and keep this issue scoped to the broader documentation alignment; 4. update both English and Chinese docs in the same PR. A focused docs-only PR against `dev` would be welcome. Please cite the source paths you align with in the PR description so reviewers can verify the doc/code mapping quickly. -- 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]
