SEZ9 commented on PR #11545: URL: https://github.com/apache/seatunnel/pull/11545#issuecomment-5806506176
Thanks for the latest round. I re-checked the open findings against what is in this thread, but I don't have the changed-file content in front of me to confirm the fixes, so I'd like to close these out with concrete evidence rather than mark them resolved on description alone. **F5 (leftover JDK 8 / cryptic `Unrecognized option`)** — could you point to the fail-fast Java-version check in the launcher scripts (or paste the relevant snippet)? **F2 (silent Kerberos reflective reload failure)** — what now detects or reports the missing `--add-opens`/`--add-exports` condition at runtime? **F4 / F7 (old 2.3.13 release under JDK 17 in `upgrade_compatibility.yml`)** — how does the old cluster start cleanly there now? A link to a passing run or the workflow change would be enough. **F1 (module flags only in user-overridable `config/jvm_*_options`)** — if the launcher now appends the flags idempotently regardless of the shipped config files, could you show that logic and add a short note in `incompatible-changes.md` so operators aren't surprised to see the flags on the command line? **F3 / F6 / F8 (blanket module opens + workflow-level `JAVA_TOOL_OPTIONS` in `backend.yml`)** — moving away from a global `JAVA_TOOL_OPTIONS` is the right direction. The thread mentions that the replacement (a JDK-activated Maven profile in the root `pom.xml`) currently breaks the default build and undercuts the JDK 17 unit-test lane's signal. Could you: 1. Paste the concrete error from a plain Maven build on the current head (JDK 11 and JDK 17) and explain what the profile activation changes? 2. Clarify how the JDK 17 lane still proves the build works on a plain JDK 17 if the profile is always active there? If the flags are only needed by specific test modules, scoping them to those modules' test configuration would keep the lane meaningful and narrow the `--add-opens` surface (F3). The thread also reports a merge conflict against `dev`; that will need resolving before this can move forward. Once the build break, the conflict, and the F3/F6/F8 mechanism are sorted out and the evidence above is in, I'm happy to take another pass. <!-- streview-comment:1270 --> -- 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]
