fix(test-runner): report Worker Cleanup step for passing tests

[vazidmansuri005]

Fixes #38350

Problem

The Worker Cleanup step (which includes fixture teardown timing) was only reported in result.steps when a test failed. For passing tests, this data was omitted from the onTestEnd reporter callback, making it impossible for custom reporters to identify slow fixture teardowns in successful runs.

Root cause: In workerMain.ts, the Worker Cleanup step was gated behind if (this._isStopped), which is only set to true on test failure. For passing tests, fixture teardown happened later in gracefullyClose() using a fake TestInfo with no-op step callbacks — so no step data was ever reported.

Fix

Extended the condition from:

if (this._isStopped) {

to:

if (this._isStopped || !nextTest) {

When nextTest is undefined (last test in the group), the Worker Cleanup step now runs and reports timing data as part of that test's result.

Performance guard: Worker fixture teardown (teardownScope('worker')) is only performed when this._isStopped is true (worker is actually shutting down). When !nextTest triggers the cleanup on a passing test, only test-scoped fixtures and afterAll hooks are processed (both no-ops at that point). This preserves worker fixture reuse across test groups — the browser is not restarted between files.

// Only teardown worker fixtures when the worker is actually stopping.
if (this._isStopped) {
  await this._fixtureRunner.teardownScope('worker', testInfo, { type: 'teardown', slot: teardownSlot });
}

Test plan

• Added new test: should report worker cleanup step for passing test
• Updated existing test expectations in test-step.spec.ts to include the now-reported Worker Cleanup step
• All 43 tests in test-step.spec.ts pass
• reporter.spec.ts (32 pass), playwright.trace.spec.ts (45 pass), playwright.connect.spec.ts (5 pass)
• Lint clean

[pavelfeldman]

Looks like AI generated blob

[vazidmansuri005]

Looks like AI generated blob

Hi @pavelfeldman, fair point — the test file changes looked like a bulk edit and I understand why that raised a flag.

The actual code change in workerMain.ts is small (extending if (this._isStopped) to if (this._isStopped || !nextTest) to trigger Worker Cleanup on the last test in a group). The
large number of test updates was a side effect of the new step appearing in every test's output that previously didn't include it.

Would you be open to guidance on the preferred approach here — should I resubmit with only targeted new tests and accept that existing tests need updating, or is there a better way
to structure the fix? Happy to rework it.