CryoThrust commented on PR #12101: URL: https://github.com/apache/seatunnel/pull/12101#issuecomment-6966828199
Pushed `841c8d50c` with a test-only strengthening for F4. Here is the current mapping: - **F8 grouping:** I checked the whole `SeaTunnelConfValidateCommand` at this head. It has no condition combining `&&` with `||`; the mixed expression is in `ClientCommandArgs` and is explicitly grouped. The command's other conditions use either operator separately. My earlier pointer was to that actual mixed expression, not a change in the command file. - **F8 duplication / F6:** `executeForExitCode()` has the single `System.out.println(result.toJson())` emission, after `validateResult()` returns. The latter's try/catch returns a success/failure model and does not serialize or print. Thus success serialization is outside validation error handling, and there are no separate failure-JSON emission blocks. - **F4:** The Bash/Zeta test already checked one stdout line and parsed `valid` for both success and missing-file failure. That parser could accept a first JSON value without rejecting trailing tokens. The new commit uses a local Jackson mapper with `FAIL_ON_TRAILING_TOKENS` for both paths, retaining the one-line assertion and exit-code checks. A second JSON value or trailing log text now fails parsing. The actual container test passes: one case, no failures/errors/skips. Native Windows remains outside this test's coverage. - **F5:** `testNullMessageAndPathResolutionPreserveTextContract` mocks `FileUtils.getConfigPath()` to throw and asserts that `execute()` throws the same `IllegalArgumentException` instance. `getConfigPath()` remains before the validation try/catch, so its type, message and cause are not wrapped or rewritten. This pins exception propagation, not every filesystem/platform error wording. - **F2:** `ConfigValidationResult.toJson()` builds Jackson `ObjectNode`/`ArrayNode` values and uses `put` for each message, then serializes the tree; it does not concatenate messages into JSON. `ConfigValidationResultTest.escapesSpecialCharactersInErrorMessages` round-trips quotes, newline and tab characters. - **F7:** Both EN/ZH `engines/zeta/user-command.md` guides describe `schemaVersion`, `valid`, `phase` and `errors`, including error-entry fields and the Bash-only coverage boundary. F1/F3 remain as implemented in `90fd7b7eb`, with the credential-stack-trace and text-mode null-message regressions named in the previous reply; this follow-up changes no production code. The [fork Build on `90fd7b7eb`](https://github.com/CryoThrust/seatunnel/actions/runs/44012181068) is now successful: 88 successful jobs and 9 conditional skips. Pub/Sub JDK 11 passes on retry ([job 134068552589](https://github.com/CryoThrust/seatunnel/actions/runs/44012181068/job/134068552589)); that does not establish the earlier cleanup wait's cause. The new test-only head still needs fresh CI. Its actual Zeta/Bash integration test, scoped formatting, E2E-module `-DskipTests verify` and whitespace checks pass locally. No timeout or assertion workaround was added to Pub/Sub or engine tests. -- 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]
