SEZ9 commented on PR #13261: URL: https://github.com/apache/seatunnel/pull/13261#issuecomment-6936140370
@SEPURI-SAI-KRISHNA Thanks for the point-by-point update. You're right that my notes referred to `24b8e5153a`, and the current head is `50e1c1087a`. I have not yet checked these points against the code at `50e1c1087a`, so I'm not marking any of them resolved from the description alone. What you describe does address what each point asked for: - **F1**: `run()`, `addSplitsBack`, `registerReader` and `assignSplit` all `synchronized`, so the check and the assignment cannot interleave, plus the concurrency test you describe. - **F2 / F5**: `testRestoredSplitIsAssignedOnceWhenTheReaderRegistersAndRunFollows` for the Zeta order (one assignment, no second `new SingleSplit(null)`), and `testSplitGivenBackWhileTheReaderIsRegisteredIsAssignedAtOnce` for `run()` after a re-assignment. - **F4**: the Javadoc stating that the reader restores its split from its own checkpoint state. - **F7**: `public` test class and methods, and the debug log on the take-back path. These stay open until they are checked against the diff. One question on **F6**, which your reply does not mention: which test covers the Zeta restore ordering where `addSplitsBack` is called before the first `run()`? If it is `testRestoredSplitIsAssignedOnceWhenTheReaderRegistersAndRunFollows`, please confirm that the call order there is `addSplitsBack` -> `registerReader` -> `run()`. <!-- streview-comment:2389 --> -- 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]
