SEZ9 commented on PR #13261: URL: https://github.com/apache/seatunnel/pull/13261#issuecomment-6869986941
Thanks for the detailed follow-up on `24b8e5153a`, @SEPURI-SAI-KRISHNA. **1. Lock (F1)** — Thanks for the explanation of the lock-order reasoning and the note that the window may already exist on dev. I'll verify the `synchronized` changes and `testReaderRegisteringDuringAnAssignmentDoesNotAssignTheSplitAgain` against the diff before closing this. One ask in the meantime: please make sure that test is deterministic (e.g. a latch or barrier rather than timing) so it won't be flaky on slow CI runners. **2. Split back while the reader is registered (F2 / F5 / F6)** — `testSplitGivenBackWhileTheReaderIsRegisteredIsAssignedAtOnce` and `testSplitGivenBackGoesToARegisteredReader` sound like they target exactly the orderings I asked about (split returned while the reader is still registered, `run()` after a re-assignment, a second reader id). I'll confirm against the diff. **4. Javadoc (F4)** — Your comment appears to be cut off at "it now sa…", so I can't see the new wording. Could you paste the final Javadoc text? What I'm looking for is a sentence on *why* the early return on an empty list is safe (the reader restores its own split from checkpoint state). **F7** — Did the commit also make the test class/methods `public` to match the module's other tests, and add a debug log on the take-back path? Both are minor. Once I have the Javadoc text and an answer on F7, I'll go through the commit and follow up. <!-- streview-comment:2325 --> -- 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]
