123123213weqw opened a new pull request, #6385: URL: https://github.com/apache/rocketmq-dashboard/pull/6385
@lizhimins Supersedes #5760 (GitHub refuses to reopen maintainer-closed PRs, hence the fresh one) — same fix, with all three review asks applied, and the branch merged onto the current `rocketmq-studio`. ## What `decide()` checked the abort reason ahead of the projected success terminal, so a stop landing *after* the success frame had already been projected (row written, `run_finished` published) rewrote the row to `STOPPED` and added a second terminal event over an answer the user watched complete. Since `successTerminalProjected` and `stopRacedSuccess` are strictly complementary, moving the projected-success check ahead of `abortReason` is exactly "an abort only wins when it preceded the success frame"; the existing stop-during-the-run case still passes. ## Review asks from #5760, applied - The test now lives in a dedicated `AiRunExecutorTest` rather than `AiRunServiceTest`. - `StubAgentProvider` gains an `afterEvent` hook (runs after each scripted frame reaches the sink, with the stream's options in hand) — the stop-in-the-middle-of-the-stream case no longer needs an anonymous provider; the test scripts one success frame and stops from the hook. - The `AiConversationProperties` construction is shared: `setUp()` and the late-executor test both go through one `properties()` helper. `AiRunExecutorTest` is green on this branch. Taken together with the re-read side of the same corruption (#5761, also resubmitted). -- 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]
