Failed
Open on GitHub

fix: use upstream Playwright cache validation and Rsbuild port support

SSheraff
80173803
8356Latest execution

Self-Healing CI

Classification: Flaky task

We classified this as a flaky task rather than a code change. The failing assertion (expect(fullPageLoad).toBe(true) in redirect.spec.ts:71) checks timing-sensitive reload detection that has a recorded flakiness rate of ~2.82%, and none of the PR's changes touch redirect behavior or the affected E2E app. With only 1 of 120 tests failing, a rerun should resolve this without any code modifications.

An empty commit was automatically applied to the branch to trigger a new CI pipeline execution to resolve the flaky task.

The AI model provider was experiencing issues when this fix was generated.

The failing test (tests/redirect.spec.ts:71) asserts that a full page reload was detected when navigating with reloadDocument: true. The assertion expect(fullPageLoad).toBe(reloadDocument) received false instead of true — meaning Playwright did not detect the reload event in time.

Key reasons this was not classified as a code change:

  1. No direct code correlation: The PR's changes are confined to Rsbuild watch-exclusion logic, Babel loader option handling in solid-start, and package version bumps (Playwright 1.61→1.63, Rsbuild 2.1→2.2.5). None of these touch redirect behavior, the reloadDocument flag, or any logic in redirect.spec.ts or its underlying application code.

  2. Project not in touched_projects: The failing project tanstack-router-e2e-solid-basic-file-based does not appear in touched_projects (which is empty), indicating no files belonging to this E2E app were modified by the PR.

  3. Established flakiness history: The task has a recorded flakiness rate of ~2.82%, confirming that this test has failed intermittently in the past without code changes. Full-page-reload detection in E2E tests is inherently timing-sensitive — a race between the browser's reload event and Playwright's listener can cause this assertion to fail non-deterministically.

  4. Isolated failure: Only 1 of 120 tests in the suite failed, which is consistent with a sporadic timing issue rather than a deterministic code regression.

  5. Playwright version change as context: While Playwright was upgraded from 1.61.0 to 1.63.0, no changelog evidence points to a behavioral change in reload detection that would deterministically break this assertion. The upgrade most plausibly shifted timing margins enough to surface an existing flaky assertion on this run.