Principles for Automated Testing

Published See discussion on Bluesky

This post is meant to be a living reference for common patterns and principles that I personally find important when working within an automated test suite.

Some of these may only apply to specific scenarios, some you may disagree with, but so far they've all proven in my own experience to be incredibly valuable to maintaining a large automated test suite for a long time and with a lot of contributors.

Creating a test should be incredibly easy

If you want to encourage that tests get written and that your code is generally well covered then it should be dead easy for any contributor to create a (well structured) test.

Test output should be "clean"

Specifically this means that logs to stdout (e.g. console.log) should either:

  • Fail the test
  • Be mocked during the test

The test output should be incredibly easy to debug failures, if there are several logs unrelated to the failure then it will be more difficult to understand how the test failed and what needs to be done to fix it.

More coming soon!