Skip to content

A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.

2.0 KB
Raw Blame

E2E Tests

E2E tests use Playwright.

Running Tests

pnpm playwright test                           # Run all e2e tests
pnpm playwright test e2e-tests/slides.spec.ts  # Run specific test
pnpm playwright test --ui                      # Interactive UI mode

Rebuild Before Taking Screenshots

Rebuild the frontend for tests that take screenshots to view the latest changes.

make fe
pnpm playwright test

Visual Regression Tests

The focused visual suite protects shared theme, typography, chrome, and layout behavior. It uses the deterministic fixture in py/visual_tokens.py.

Run the visual suite

Visual snapshots are captured in the same pinned Playwright container locally and in CI. On macOS, the script uses the Docker CLI with OrbStack or another Docker-compatible runtime.

From the repository root, build the frontend and run the focused suite:

make visual-test

Update visual baselines

Generate committed baselines with the pinned container, not with native Playwright. From the repository root, run:

make visual-update

Review every changed PNG. Then run the suite again without --update-snapshots.

Add a visual assertion

Add a visual assertion for a shared semantic token, theme behavior, typography, product chrome, or responsive layout.

If pixels do not express the requirement, use a DOM or behavior assertion. Do not add screenshots to every functional E2E test.

If the existing fixture contains the affected semantic role, reuse it. Add a fixture state only for a named planned change.

Review a baseline change

Use the Playwright report to compare the expected, actual, and difference images. Make sure that each changed region matches the stated product change.

Never use --update-snapshots to silence an unexplained failure. Correct the fault or unstable state before a baseline update.

Make sure that unrelated baselines stay unchanged. State the visual reason for each baseline change in the pull-request summary.