SEZ9 commented on PR #12129: URL: https://github.com/apache/seatunnel/pull/12129#issuecomment-5578118319
@DanielLeens thanks for tracing the actual decode sites — that's exactly the confirmation I was after on item 3. Agreed on all counts: the `.ordinal()` encode in `GetJobStatusOperation.java:81` and the `JobStatus.values()[jobStatusOrdinal]` decodes at `ClientJobProxy.java:154` and `JobClient.java:121` are pre-existing, not introduced here, and the mixed-version `ArrayIndexOutOfBoundsException` / wrong-constant exposure is a genuinely separate RPC design item. Folding it into this `EnumSet` refactor would blow up the scope, so keeping it out is the right call. The concrete outcome I want from that is purely wording: the `JobStatusTest` pinning-test Javadoc should state plainly that the pinned ordinal table only guards same-build code, and that a rolling-upgrade Zeta cluster with a reordered or resized `JobStatus` can still fail at those decode sites. That way nobody later reads the `EnumSet` change as having closed the RPC gap. That same edit is a good moment to fix the confusing sentence in that Javadoc that I flagged earlier. So the remaining asks on my side are unchanged: - Javadoc boundary wording in `JobStatusTest` as above (plus the sentence cleanup). - Confirmation from the author on items 1/2 you referenced. - Enable Actions on the fork and push the retrigger so we get a real `Build` run instead of the "Unable to detect the workflow run" signature. Once those land I'll do the fresh pass. Nothing further needed from you until then — appreciate the thorough check. <!-- streview-comment:884 --> -- 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]
