SEZ9 commented on PR #11885:
URL: https://github.com/apache/seatunnel/pull/11885#issuecomment-6137847793

   Thanks for keeping the branch current. Since `befad345c1` the head has only 
moved through two `dev` merges (`b0ab23fb03`, `99d1db7934`), and the latest 
re-review of `99d1db7934ce` reports the PR's own added/removed lines as 
unchanged. To close out my earlier points efficiently, could you point me to 
where each of these is handled in the current diff?
   
   1. **Checkpoint round-trip for `stopOffset` (F1 / F6)** — the exact code 
that persists and restores the resolved offset in `IncrementalPhaseState`, plus 
the `addSplits()` fallback for checkpoints taken before this field existed. If 
`IncrementalSplitAssignerTest` already exercises the restore through the real 
state serializer rather than constructing the state object directly, a pointer 
to that test would settle this.
   
   2. **Stop-boundary consistency (F2)** — how a split created before snapshot 
completion and the checkpointed `resolvedStopOffset` are kept from disagreeing, 
and how multi-split assignments avoid straddling the resolution.
   
   3. **E2E readiness gate (F3)** — what the `initial`-startup `latest-stop` 
assertion now gates on to prove the snapshot is still in progress, since id=1 
reaching the sink alone does not establish that.
   
   4. **Retry on the offset lookup (F5)** — where 
`resolveLatestStopOffsetWithRetry()` lives and what its bounds are.
   
   5. **Docs and PR description (F4 / F7 / F8)** — where the `stop.mode = 
latest` semantics are documented, and a quick confirmation that the PR 
description reflects the enumerator-side implementation and the `initial`-only 
scope of the post-snapshot e2e assertion rather than `MySqlBinlogFetchTask` / 
`BoundedMySqlStreamingChangeEventSource`.
   
   Once I have those pointers I'll do a focused re-check of just these items.
   
   <!-- streview-comment:1720 -->


-- 
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]

Reply via email to