DanielLeens commented on PR #11780: URL: https://github.com/apache/seatunnel/pull/11780#issuecomment-5370790615
Confirming @SEZ9's read here — that's accurate. `codex/fix-pr11780-resolver-gate-20260820` @ `01fbe370` is a real branch on my fork; I pushed it on 2026-08-20 while root-causing the `all-connectors-it-1`/`all-connectors-it-6` failures I wrote up above. The change is a 3-line guard in `IncrementalSourceReader#initializedState`: it now only sets `restoredCheckpointTables` (the flag that later drives `output.restoreSchema(...)` on checkpoint) when `debeziumDeserializationSchema.getSchemaChangeResolver() != null`. That keeps the collector-side restore behind the same condition `SeaTunnelRowDebeziumDeserializeSchema#restoreCheckpointProducedType` already uses internally, so dialects without a resolver (OpenGauss, and any others that don't implement one) no longer get pulled back to a stale checkpoint schema on restore. I also re-verified the `75ea8e5`→`cfe2ab9` diff on the files this PR actually touches: `IncrementalSourceReader.java`'s `restoreSchema` call site is byte-identical across the merge, and the only real change in `SeaTunnelRowDebeziumDeserializeSchema.java` is the unrelated `AlterTableCommentEvent` handling pulled in from `dev`. So yes — `cfe2ab9` is a clean conflict-resolution merge, and the regression I flagged is still live on the current head exactly as described. @dybyte — once you rebase onto latest `dev`, feel free to cherry-pick `01fbe370236aadd358ec7ee99f5f9df610479e3d` (or fold the same guard in by hand) onto the refreshed head. I'll do the next review pass as soon as that's up, and we can confirm `all-connectors-it-1`/`all-connectors-it-6` go green from there. -- 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]
