We classified this failure as an environment issue rather than a code regression. The E2E server could not start because port 38873 was already occupied by a previous process (EADDRINUSE), causing all three Playwright tests to fail as a cascade. The PR only touches @tanstack/router-core internals and has no connection to the E2E test infrastructure, so a re-run on a clean environment should resolve this.
Failed after
MSManuel SchillerView latestperf(router-core): allocate replaceEqualDeep copies lazily
8362c955f6fOpen on GitHubSelf-Healing CI
Classification: Environment state
An empty commit was applied to the branch to trigger a new CI pipeline execution to resolve the environment issue.
The AI model provider was experiencing issues when this fix was generated.
The failing task tanstack-react-start-e2e-website:test:e2e was classified as environment_state rather than code_change for the following reasons:
The error is a port conflict, not a code failure. The root cause is EADDRINUSE: address already in use :::38873. The test web server (srvx) could not bind to the assigned port because another process was already holding it. All three Playwright tests then failed as a cascade — the first timed out waiting for a DOM element that never rendered, and the other two received ERR_CONNECTION_REFUSED when trying to navigate to the server.
The failing project was not touched by the PR. The PR exclusively modifies @tanstack/router-core (a lazy-allocation performance refactor of replaceEqualDeep). The failing project tanstack-react-start-e2e-website is not in the touched_projects list (["root", "@tanstack/router-core"]). There is no plausible path by which a pure algorithmic change inside replaceEqualDeep could cause a port binding failure in an unrelated E2E app.
Flakiness rate is 0%, ruling out a flaky classification. A 0% flakiness rate means this task does not exhibit non-deterministic behavior historically, so it cannot be reclassified as flaky_task.
No similar failures exist on main. The similar-task-failure detector returned no matching failures on the main branch, confirming this is not a pre-existing widespread regression — it is an environment cleanup issue (a leftover process occupying the port) specific to this CI run.