I’ve been chatting with a bunch of people lately about testing. Everything from helping people get started from square one, companies transitioning suites into new contexts (say, after moving from a monolith to a microservices architecture), all the way to teams struggling with large slow test suites that they’ve grown to hate.
One topic that comes up often is the role of mock objects in testing, particularly when practicing test-driven development. There are a lot of forceful generalizations out there (“beware of over-mocking!”, “only mock what you own!”, “only mock external systems you don’t own!"), but taken alone they don’t do very much to explain when test doubles should be used and when they shouldn’t. I have a pretty good idea why that is.
The reason is that there are two schools-of-thought when it comes to TDD as a productivity workflow. On one hand: Detroit-school TDD, the simpler red-green-refactor workflow. On the other: London-school TDD, which makes use of test doubles to direct the design of systems.
This video series explores the lesser known latter school, using a technique that I’ve come to call Discovery Testing. We’ll start with the background needed to make sense of why the tooling and terminology we use about testing is so overloaded, then use an ambitious approach to Conway’s Game of Life as an example problem to work through together, one test at a time.
There are four episodes in this series:
Some links of things referenced in the series so far: