Skip to content

description: Structure Playwright VR tests using fixtures, parallelism controls, and project metadata parameterization. tldr: Use test.extend fixtures for shared auth and page state instead of globals or beforeEach chains. Set fullyParallel: true globally but apply mode: 'serial' only to tests that share Drupal editorial state.

Test Organization

When to Use

Use this when structuring tests for shared authentication, controlling parallelism, or parameterizing tests across viewports.

Decision

Need Pattern
Shared auth / page state test.extend fixture
Shared navigation test.beforeEach
Tests that must run in sequence test.describe.configure({ mode: 'serial' })
Max throughput fullyParallel: true in config
Per-test metadata from project testInfo.project.metadata

Pattern

Fixtures (test.extend)

import { test as base, expect, type Page } from '@playwright/test';

type Fixtures = { authedPage: Page };

export const test = base.extend<Fixtures>({
  authedPage: async ({ page, context }, use) => {
    await context.addCookies([
      { name: 'SESS', value: '...', url: 'https://mysite.ddev.site' },
    ]);
    await use(page);
  },
});

beforeEach for navigation

test.beforeEach(async ({ page }) => {
  await page.goto('/');
  await page.evaluate(() => document.fonts.ready);
});

Parallelism

test.describe.configure({ mode: 'parallel' }); // tests in this block run in parallel
test.describe.configure({ mode: 'serial' });   // sequential, abort on first failure

At config level: fullyParallel: true makes everything parallel by default.

Parameterized via project metadata

projects: [
  { name: 'desktop',
    use: { viewport: { width: 1440, height: 900 } },
    metadata: { device: 'desktop' } },
  { name: 'mobile',
    use: { viewport: { width: 375, height: 667 } },
    metadata: { device: 'mobile' } },
]

Read inside tests via testInfo.project.metadata.

Common Mistakes

  • Wrong: Module-level globals for shared state → Right: fixtures are the proper Playwright alternative
  • Wrong: serial mode as the default → Right: defeats parallelism; use only when tests genuinely depend on ordering
  • Wrong: fullyParallel: true on tests that share editorial state in Drupal → Right: isolate per-worker or use serial

See Also