Why should I unit test my code?

Why should I unit test my code?

Hi everyone, today we are going to explore the WHY question on unit testing the code.

A very popular debate which I have observed in the developers’ world is whether or not to write unit test cases for code and how writing those test cases could (no, it shouldn’t) be an overhead for the developers.

In my experience of almost 7 years of coding(mostly web development), I have met very few people who willingly unit test almost everything they code. So I feel that it’s the need of the hour to talk more about it and focus more on TDD (Test Driven Development).

The more code you write without testing, the more paths you have to check for errors

As quoted by one of the articles here,

unit tests are so important that they should be a first-class language construct.

Me: Yes, please.. it’s about time now!!

What is unit testing?

So, before you shall ask me, why should I unit test my code, 🤔 let’s start by understanding what exactly is unit testing…

There are many good explanations and introductory articles present on the web which cover this topic.

As Wikipedia says:

Unit testing is a software testing method by which individual units of source code, sets of one or more computer program modules together with associated control data, usage procedures, and operating procedures, are tested to determine whether they are fit for use.

Here, the unit can be the smallest part of the application/module that can be tested in isolation. It can be a class in object-oriented programming, a function in procedural programming, a component in React, Angular, Vue, and so on.

The main purpose is to verify if that unit of code is working fine for all the scenarios which might occur. If yes, then whichever module consumes it, we are sure that this code will not break.

Unit testing makes the process more agile. Ask me how?

This is because as the application evolves we need to add more and more features, and some times it involves revisiting earlier designs and code.

Unit tests can detect breaking changes in design contracts, and issues can be found very early and resolved.

They also help in safer refactoring of the code. If we are refactoring code, the end purpose of that code should remain the same, and if we have proper unit tests in place, any defect in refactored code can be easily detected by the test cases.

So we know where to debug if any of the test cases fail. This also makes the debugging process easier.

We have plenty of reasons to unit test the code, but now enough of talks, let’s try to explore the importance by coding it out.

Note: All my examples below are written in JavaScript and the unit test framework used is Jest.

Example 1:

Let say, we want to write a function which gets two numbers as input, adds the and returns the result, as below:

function addNumbers(num1, num2) {
   return num1 + num2;

And, we will write test cases for the above function. PFB the code for it:

expect(addNumbers(5, 6)).toBe(11);
expect(addNumbers(-1, 9)).toBe(8);

Till now, everything looks fine. We can pass two numbers and we get the desired result.

Next, imagine a situation where the consuming module forgot to pass parameters to addNumbers function. What should happen then?

Let’s say I want it to return 0 when any unexpected (undefined, null, string, etc.) value comes in. The current code will break because it will not return 0 when any unexpected inputs are passed to it.

Let's now go ahead and add more tests to handle these scenarios.

// fails as with current implementation, result will be NaN

// fails as with current implementation, result will be 'abcxyz' expect(addNumbers('abc', 'xyz')).toBe(0);

When we run our test, these two test cases will break, since the actual result is different than what we had expected from this method.

So, we know that the code needs some modification to work properly and we go ahead and update the implementation, as below:

function addNumbers(num1, num2) {
   return parseFloat(num1 + num2) || 0;

Here, parseFloat will return NaN if we pass any non-number input and since NaN is a falsy value in JavaScript, the function will return 0 for these scenarios.

Now, all our test cases should pass, and if we had delivered this code today to production, we can sleep peacefully because we know that this method will not break any module from which it is getting consumed. What a relief, isn’t it 😎

Example 2:

Let’s consider we had a method which takes in userDetails as its argument and returns a new object, as below:

function getUserDetailsToDisplay(userDetails) {
   return {
     name: getFullName(userDetails.firstName, userDetails.lastName),
     dateOfBirth: userDetails.dob || '-'
function getFullName(firstName = '', lastName = '') {
   return (firstName + ' ' + lastName).trim() || '-'; 

And the corresponding test case goes as below:

const resultWithValidDetails = getUserDetailsToDisplay({
   firstName: 'testFirstName',
   lastName: 'testLastName', 
   dob: 'test date'
expect(resultWithValidDetails.name).toBe('testFirstName testLastName');
expect(resultWithValidDetails.dateOfBirth).toBe('test date');
const resultWithNoDetails = getUserDetailsToDisplay({});

Next, a new requirement comes in, wherein, we need to show the middleName of the user as well. So we go and change the functionality of function getUserDetailsToDisplay as below:

function getUserDetailsToDisplay(userDetails) {
   return {
     name: getFullName(userDetails.firstName,     
           userDetails.middleName, userDetails.lastName),
     dateOfBirth: userDetails.dob || '-'

At this point, let say that we forgot to update the getFullName , and we also ignored any warning that we got. So If we had the test cases in place, they would start breaking, and this will alert us to examine the code in getFullName.

PFB the exact error we will get while running the tests:

Error while running above test case which shows that expected and received objects are not same

Seeing this error, we understand there is an issue with our code, and we know exactly what is breaking. So, we could now go and fix the logic of computing names.

There could be a second scenario, where, if we are following TDD, we would have already updated the test cases to test the new changes as below:

const resultWithValidDetails = getUserDetailsToDisplay({
   firstName: 'testFirstName',
   middleName: 'testMiddleName'
   lastName: 'testLastName', 
   dob: 'test date'
expect(resultWithValidDetails.name).toBe('testFirstName testMiddleName testLastName');
expect(resultWithValidDetails.dateOfBirth).toBe('test date');

But, these test cases would still fail, because the getFullName expects three parameters now, but we are only passing 2 of them in actual code. This is an example of easier debugging, we know where and what code block to debug based on the unit test case failed. PFB error in this case:

Snapshot of error for above use-case

Looking at error, again, we found out the mistake and could easily fix the code, let say, as below:

function getFullName(firstName = '', middleName = '', lastName = '') {
   return (firstName + ' ' + middleName + ' ' + lastName).trim() || 

Now, all test cases should pass, and we should be good to go. So far, by writing unit tests, we have prevented two bugs in production, thereby reducing the cost of delivery (by avoiding the extra iteration of finding bug, fixing it, verifying it, and delivering it again to production).

The quality of the code gets better when it is tested properly, and hence quite less prone to breaking our application/product.

By now, at least you have some idea about how unit testing might help. Now, I know these might not be one of the greatest examples out there, but I purposefully kept it so simple, so that we are comfortable with the very idea of unit testing.


If you ask of my take on it, unit tests should not be seen as an overhead or burden, instead, we should embrace them as a daily practice and a healthy coding habit.

There have been many times where I have thanked my past self for writing that piece of test case which saved me from making a huge mistake and preventing an otherwise almost certain production bug.

Unit tests help us write better code and handle the multiple real-time scenarios quite well.

I always stress on the point to write logical test cases, and not to write it just for the sake of achieving that coverage goal, so that if anyone makes any design change in the component/function, the unit test should fail and that’s how any developer, new to that codebase, can understand what was the original purpose of that code.

This covers one more purpose of unit test, which is documentation of the code. Good unit test cases serve as documentation for future and it’s easy to maintain a codebase that is well documented and properly unit tested.

PFB one quote I once read on the web (I do not remember the author's name, but they must be a legend).

Any code that is not unit tested is legacy code.

So, if you want to ask me when should be the right time to start unit testing my code, the answer would be NOW!

Note: This article is the first of three articles I have on unit testing. PFB the series topics:

  1. Why should I unit test my code? (this article)
  2. How should I unit test my code? (Setup and examples)
  3. What should I unit test in my code?

Stay tuned and catch you in the next article !! Any feedback is highly appreciable.

Thanks for reading. Stay safe, keep on learning and have a great day!!