SEZ9 commented on PR #11983: URL: https://github.com/apache/seatunnel/pull/11983#issuecomment-5433990021
Thanks for the review — the three points were all correct, including the one that caught a factual error in my own PR description. All are fixed. **Released-behaviour change, missing `incompatible-changes.md` entry.** You were right and my PR body was wrong. I claimed this was unreleased-branch behaviour; I re-checked and `POST /hazelcast/rest/maps/log-level` was added in 8241ec193 (#3025) and has shipped in every release since 2.3.0, so a caller sending a mis-cased or unknown level does see a `200`/`SUCCESS` turn into a `400`. Added an entry under "API Changes" in `docs/en|zh/introduction/concepts/incompatible-changes.md` (affected component, description, impact, migration guide) and corrected the PR description. **Non-deterministic level list in the error message.** Correct: `Level.values()` returns the values of a `ConcurrentHashMap`, so the "valid levels are: ..." list could come out in a different order on each run. `validNames()` now sorts; `Level implements Comparable` on `intLevel`, so the natural order is most-severe-first (`OFF, FATAL, ERROR, WARN, INFO, DEBUG, TRACE, ALL`). Covered by a new test that asserts relative order via `indexOf` rather than exact string equality, so a custom level registered by another test in the same surefire JVM cannot break it. **Wrong javadoc on `parse`.** Also correct — the javadoc said `Configurator` "silently ignores" a `null` level, which is the opposite of what happens: it clears the explicit level, so the logger falls back to its parent and the root logger falls back to `ERROR`. That is the whole reason this bug silently *lowered* levels rather than doing nothing. The javadoc and a stale comment in the test now say that. The Windows unit-test red is not from this change: it is the known `AbstractSeaTunnelServerTest.before:70` "Node failed to start!" Hazelcast-on-Windows flake in `TaskTest`. `LogLevelsTest` is 15/15 green in that same 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]
