Good Practices for Writing Tests¶

Estimated time to read: 2 minutes

For our tests to be useful we need them to have certain characteristics.

Fast. Nobody will be happy to run them if they take ages to execute. You should want to run them for every single change.
Simple. We don't want to write tests for our tests, hence it should be easy to spot if a test is implemented correctly or not.
Focused. If a test fails it should be clear what broke exactly. You don't want to start analysing what you need to fix, especially when you're in a hurry.
Deterministic. The same test run over the same business code should yield the same result again. Nobody wants to run a test suite that is unpredictable or flaky.

Test Implementation¶

A single test should do a single thing. If you want to verify 3 aspects of your code write 3 tests.
Make your test code readable and easy to understand. When you come back to update your test code it should all be self-explanatory.
With unit tests, test your own code only. Avoid driving a Web browser, a database or an external API. Restructure your code or mock those dependencies away.

Support your test suite by QA tools, such as linters. They will spot syntax errors, code smells and other issues even before your test code unveils them. You can also run them over your own test code.
Run your tests automatically. Locally, e.g. configured in your IDE, and in your CI/CD pipeline, every time you push a change.