DanielLeens commented on PR #11780: URL: https://github.com/apache/seatunnel/pull/11780#issuecomment-5412606712
Thanks for the summary, @SEZ9. On "your review comment appears truncated mid-sentence at 'Issue 2 (carried over from 202'" — I just re-fetched that review directly from the API (`pulls/11780/reviews/5004128771`, 18909 bytes) rather than relying on the rendered page, and it is not truncated. The full sentence, unedited, is: > Issue 2 (carried over from 2026-08-13/19/20/23, still High, still blocking) — confirmed still unaddressed, and still reproducing on this exact head's CI. I re-read `IncrementalSourceReader.java` at the current head directly (not from memory of the prior diff) to make sure the one new commit didn't incidentally touch this path — it didn't. `restoreCheckpointState(...)` (`IncrementalSourceReader.java:253-287`, the same method the legacy-checkpoint-restore commits added in the previous round) still unconditionally returns whichever `checkpointTables` it resolved — from the new-format field, or from `restoreLegacyCheckpointTables(...)` — with no check on whether the deserializer's `SchemaChangeResolver` is non-null: [...] It continues from there into the rest of the analysis and the CI section quoted in your own summary (the `380d77ae19` / fork run `32637942572` breakdown), so nothing is missing on the thread — this was a "205" vs "2026" character read, not an actual cutoff. Same false-truncation pattern I just had to correct on PR #11809's thread a few minutes ago, for what it's worth. On substance, no disagreement — your F1-F8 summary matches my last review exactly, and I confirm the live head is unchanged at `380d77ae19` since that review, so there's nothing new to re-verify: Issue 2 (the missing `SchemaChangeResolver != null` gate on the collector-side restore) remains the sole blocking item, the fix has been independently verified three times now (2026-08-19, 2026-08-20, 2026-08-21) on `codex/fix-pr11780-resolver-gate-20260820` @ `01fbe370`, and it hasn't landed on this branch's own head yet. F2-F8 are accurately carried forward as non-blocking/secondary. Nothing to do on my side until a new commit lands. Once Issue 2's gate (and ideally F2-F3) are folded into this branch's head, I'll do a fresh pass and re-check CI on the actual fork run. -- 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]
