Write tests. Not too many. Mostly integration.
October 16, 2017
Guillermo Rauch tweeted this a while back. Let's take a quick dive into what it means.
Current Available Translations:
I've given this blog post as a talk which you can watch here:
Write tests. Not too many. Mostly integration.— Guillermo ▲ (@rauchg) December 10, 2016
Write tests. Not too many. Mostly integration.
This is deep, albeit short, so let's dive in:
Yes, for most projects you should write automated tests. You should if you value your time anyway. Much better to catch a bug locally from the tests than getting a call at 2:00 in the morning and fix it then. Often I find myself saving time when I put time in to write tests. It may or may not take longer to implement what I'm building, but I (and others) will almost definitely save time maintaining it.
The thing you should be thinking about when writing tests is how much confidence they bring you that your project is free of bugs. Static typing and linting tools like TypeScript and ESLint can get you a remarkable amount of confidence, and if you're not using these tools I highly suggest you give them a look. That said, even a strongly typed language should have tests. Typing and linting can't ensure your business logic is free of bugs. So you can still seriously increase your confidence with a good test suite.
Not too many.
I've heard managers and teams mandating 100% code coverage for applications. That's a really bad idea. The problem is that you get diminishing returns on your tests as the coverage increases much beyond 70% (I made that number up... no science there). Why is that? Well, when you strive for 100% all the time, you find yourself spending time testing things that really don't need to be tested. Things that really have no logic in them at all (so any bugs could be caught by ESLint and Flow). Maintaining tests like this actually really slow you and your team down.
You may also find yourself testing implementation details just so you can make sure you get that one line of code that's hard to reproduce in a test environment. You really want to avoid testing implementation details because it doesn't give you very much confidence that your application is working and it slows you down when refactoring. You should very rarely have to change tests when you refactor code.
I should mention that almost all of my open source projects have 100% code coverage. This is because most of my open source projects are smaller libraries and tools that are reusable in many different situations (a breakage could lead to a serious problem in a lot of consuming projects) and they're relatively easy to get 100% code coverage on anyway.
There are all sorts of different types of testing (check out my 5 minute talk about it at Fluent Conf: "What we can learn about testing from the wheel"). They each have trade-offs. The three most common forms of testing we're talking about when we talk of automated testing are: Unit, Integration, and End to End.
As indicated here, the pyramid shows from bottom to top: Unit, Integration, E2E. As you move up the pyramid the tests get slower to write/run and more expensive (in terms of time and resources) to run/maintain. It's meant to indicate that you should spend more of your time on unit tests due to these factors.
One thing that it doesn't show though is that as you move up the pyramid, the confidence quotient of each form of testing increases. You get more bang for your buck. So while E2E tests may be slower and more expensive than unit tests, they bring you much more confidence that your application is working as intended.
This is why I created "The Testing Trophy" 🏆
"The Testing Trophy" 🏆— Kent C. Dodds (@kentcdodds) February 6, 2018
- End to end w/ @Cypress_io ⚫️
- Integration & Unit w/ @fbjest 🃏
- Static w/ @flowtype 𝙁 and @geteslint ⬣ pic.twitter.com/kPBC6yVxSA
Here's another illustration of the importance of integration tests:
Still love this one. Unit testers be like: "Looks like it's working" pic.twitter.com/KiNT4wXP4a— Kent C. Dodds (@kentcdodds) August 4, 2015
We've written unit tests that verify the man can run in place, do pushups, and
read, but the man isn't integrating with his various body parts very
effectively. It doesn't matter if your button component calls the
handler if that handler doesn't make the right request with the right data! So
while having some unit tests to verify these pieces work in isolation isn't a
bad thing, it doesn't do you any good if you don't also verify that they
Integration tests strike a great balance on the trade-offs between confidence and speed/expense. This is why it's advisable to spend most (not all, mind you) of your effort there.