weiqingy commented on PR #28973: URL: https://github.com/apache/flink/pull/28973#issuecomment-6022811565
PR-1 and PR-2 have merged, so this is rebased onto master and ready for review. The diff is now this step alone. This is the first PR in the stack where the feature is observable, so the part I would most value a second opinion on is where the opt-in flag is allowed to be set, under "Why the flag is set where it is" above. The claim is that a serializer can carry the flag only on a path that actually invokes `migrate`, and that everything else leaves it off and rejects the restore rather than accepting a migration nobody performs. Does that boundary hold where you would expect it to? One scope note I have corrected in the description: `ListState<RowData>` and `MapState<K, RowData>` are in FLIP scope rather than follow-up work. They are armed in PR-3b, which extends this same seam by one structural level. The landing table now shows that step. @gyfora @RocMarshal, would you have time to take a look? -- 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]
