Playwright, didn't know I missed it

Introduction

For a recent project I was introduced to Playwright. It promises reliable end-to-end testing for web applications. After initially being skeptical (due to experiences with other test automation tools), I was soon amazed by the ease of use and stability.

In a nutshell

The installation worked flawlessly and the interactive running mode (playwright test --ui) provides immediate insight into how a test runs. The interactive generation of code for tests (playwright codegen) also works very well without generating bloat.

A little caveat

At first I didn’t realize that:

• actions are async (and must therefore be awaited)
• locators are not async
• locators are lazy

This implies that the following assertions can both be true if you are testing a dynamic page:¹

await page.goto('/')
const sections = page.locator('section')

await expect(sections).toHaveCount(1)
await expect(sections).toHaveCount(0)

The above test will pass if there is initially a single section on the page and it is later removed by the tested application. I was surprised!

Conclusion

All in all, Playwright is a pleasant discovery for me!

Notes