loading...

Test-driven development in JavaScript

tomekbuszewski profile image Tomek Buszewski ・3 min read

Testing the code we are writing is crucial in the job. Even though there are teams that doesn't do tests at all, this is one of the most important parts of successful delivery.

There are many approaches to testing software. One of my favorite is TDD, short for test-driven development. The reason it stands out is, it inverts the natural (or so it seems) flow of writing first the logic and then the test. This approach is, first the test, then the logic.

Why TDD makes sense

At first, this may seem strange. Why test something that isn't working? Why check the obvious? Think differently, think of setting requirements and assumptions for your code. When you get a task, it compels you to break it in the smallest pieces possible and write assumptions for it.

Take a Fibonacci sequence generator, for example. The goal is to create a function that will accept one parameter and will return an array of numbers. Pretty simple. What should we test?

  • it should throw an error for parameter other than integer;
  • it should return an array;
  • it should throw an error when parameter is 0;
  • it should return [0] when parameter is 1;
  • it should return an array with integers in it;
  • it should return [0, 1, 1, 2] with parameter being 4.

Take a look at that list. Six cases. Yes, six cases, not six lines of text. These are easy transferrable to a test. Observe:

it ("should return an array", () => {
  expect(Array.isArray(fib(5))).toBeTruthy();
});

This notation is super simple and allows to plan in advance.

The three cycles of test-driven development

One of the most important things in TDD is to create a cycle for yourself. It consists of three stages – red, green and refactor.

  • the red stage is writing the test and running it. It will fail and the IDE or any other runner will be red;
  • the green stage is writing the code for given test. It has to pass, but doesn't have to be well written;
  • the refactor stage is meant to get your code up to the standards;

By the end of the cycle your fraction of code should be tested and coded with current (project) standards in mind. Keep in mind that those cycles should be similar, if not the same, in length. Test-driven development works nice with the Pomodoro technique.

How can this be presented? Let's try to write a case for returning an array.

First, we create a test (red):

// index.test.js
const fib = require(".");

describe("fib tests", () => {
  it("should return an array", () => {
    expect(Array.isArray(fib(5))).toBeTruthy();
  });
});

Running it will fail, probably because we don't even have a index.js file, or if we do – it doesn't have any content.

Let's start the green phase.

// index.js
const fib = (target) => {
  const collection = [];

  while (collection.length < target) {
    collection.push(null);
  }

  return collection;
};

This code works, running the test now will turn out fine, meaning it meets the assumptions.

But, using while loop seems a little bit smelly. Perhaps we should use functional paradigm and have a recursion! Let's refactor:

const fib = (target, col = [0, 1]) => {
  if (col.length === target) {
    return col;
  }

  const newCollection = const newCollection = [...col, null];

  return fib(target, newCollection);
};

Result didn't change, but this code looks better. (I know I should make use of TCO, but I didn't want to obscure the picture).

I won't write further tests here, you are free to do so by yourself. And you can check your results or get a helping hand in my Codesandbox.

Conclusion

I have shown here the basic usage of test-driven development. It does give you the glimpse of how this technique works and what benefits it brings. But to really appreciate it, you should work with it for some time. And I strongly encourage you to do so!

Entire code and tests on Codesandbox

Posted on Mar 2 '19 by:

tomekbuszewski profile

Tomek Buszewski

@tomekbuszewski

Tech lead, recently mostly front-end.

Discussion

markdown guide
 

That's exactly what I'm thinking when I code. Everything should work or fail as expected.

When I make big reports I test a lot, specially with dates and SQL and verifying groups of records.

One of our task it's try to predict when something could fail. With experience you become better at it. Since I work with a programmer team I want to know more about testing automation and QA