Confidence in shipping a Drupal-based product shouldn't take 6 days of GitHub runner time, only to be greeted by flaky tests in some of our 500+ jobs. This is the story of how we migrated from Behat and PHPUnit to custom tooling on top of Playwright to reclaim velocity and confidence in our testing suite.

Prerequisite

Basic programming understanding and a familiarity with testing in PHP will help you get the most out of this session.

Target Audience

A medior developer with an interest in testing or a QA person who is responsible for continuous integration

Outline

Every project accumulates complexity. After 10 years, Open Social's test suite demanded 500 CI runners and hours of waiting, sometimes stumbling on flaky tests. This is the story of how we revisited that foundation and rebuilt our testing tooling on top of Playwright to reclaim velocity and confidence.

In this session I'll use the testing pyramid to audit a test suite: what are you testing, at what cost, is the confidence worth it.

From there, I'll explain the three concrete building blocks of our new tooling:

QA API or Test Control Interface (TCI) – a hidden, environment-gated HTTP API inside Drupal that provides attribute-driven PHP operations. Modules declare QA operations; the TCI publishes them as an OpenAPI spec. This decouples test data setup from the test runner entirely and allows tooling to reach into your product in a version agnostic way.

Playwright as the harness – unlocks first-class trace files, automatic retries, and parallel execution that actually scales. We'll cover how Playwright's test insights changed how we debug failures.

Human-readable specs – Using the TCI, Cucumber makes UI tests readable by non-engineers and increasingly useful as context for AI tooling. For GraphQL, a new Markdown-based spec format lets you embed queries, variables, and expectations in a file that doubles as documentation.

You'll leave with a concrete mental model and a replicable architecture for rebuilding test confidence without rebuilding your entire stack at once.

Learning Objectives

You'll learn to critically evaluate your testing suite by and thinking about tests according to the testing pyramid.
You'll learn about Playwright's amazing testing insights for integration and end-to-end testing to help solve bugs sooner.
You'll learn about how we use Cucumber and Markdown to create human readable UI and API tests on top of Playwright.

Experience level
Intermediate