So you Googled "should use var or let const es6 best practice good bad", and read enough Medium articles to feel certain you should always use const ...or maybe it was let. Either way, if an article du jour left you wondering whether var was silently removed from all the browser engines, there's a good chance You Don't Know JavaScriptâ„¢.

Let's talk about var, the issues let and const came to solve, and maybe get a better idea of when and where each of these should be used.

  1. The problem with "var"
  2. "let" examples
    1. The "for" loop
    2. The "if" block
    3. The late-comer example
  3. "const" examples
  4. Hoisting

The problem with "var"

JavaScript has always been painfully forgiving with its type casting, so it's on the developer to leave a trail of clues for themselves and future devs about the type and restrictions of any object they create.

Here are two examples of those clues:

  1. ALL_CAPS_UNDERSCORING naming convention for values that are relied on to be unchanging. These objects are called constants.
  2. Manually hoisting variable declarations to show that their later, block-scoped assignments are temporary; we let them have different values sometimes.

The common principle of any of these "clues" is hoisting. You can use any variable declaration trick in the book, but if a browser is allowed to hoist an undefined version of it to the top of your function, there's going to be trouble. Annoying, hard to debug trouble.

"let" examples

The "for" loop

The classic example of misunderstood scoping is declaring your index variable in a for loop's argument.

  for (var x = 0; x < someArray.length; x++) {
    console.log(x) // 0, 1, 2 ...

9 times out of 10, a for loop for will handle this okay, but on that 10th time you'll spend hours trying to figure out why foo is set to 5 six times.

The correct way to write a for loop is to first declare your indexer, preferably in a way that communicates the type you're going to let it be later:

  var x = 0;

for (x = 0; x < someArray.length; x++) {
    console.log(x) // 0, 1, 2 ...

With let this translates very nicely to:

  for (let x = 0; x < someArray.length; x++) {
    console.log(x) // 0, 1, 2 ...

This is why let is a thing. There are circumstances when a temporary variable is needed and you don't necessarily want extra variable declaration bloat at the top of your function. There's just no reason to be globally aware of this one-off indexer.

The "if" block

This is not a thing. I've seen this touted as a use case for let, but this is just enabling and often teaching an anti-pattern. It's a hard to read, easyily broken convention.

  console.log(thing2); // undefined

if (thing1) {
    var thing2 = 'abc';

You might be surprised to find that thing2 is undefined when it's invoked before it's even declared. This is because this code runs in browsers as:

  var thing2; // initialized as undefined

if (thing1) {
    thing2 = 'abc';

let cures this confusing bug by forcing an uninitialized var declaration which is less forgiving and causes a ReferenceError. However, from a code readability stand-point, both approaches aren't telling a very good story.

There is a way to provide the exact same outcome and keep your function's core responsibilities on its main line. This example is covered in the very next section.

The late-comer example

Sometimes you don't want to create a variable until you've done something else. My favorite example of this is the guard clause:

  if (!thing1) {

var thing2 = 'abc';

Good on ya! Creating any object in JS takes up precious run time. Why create one if there's a chance this function won't even use it? Unfortunately, browsers will turn your good intentions into this:

  var thing2; // this is initialized as undefined

if (!thing1) {

thing2 = 'abc';

The thing you were hoping to avoid happens anyway. Here's where let swoops in to save the day:

  if (!thing1) {

let thing2 = 'abc';

With let, your post-guard-clause variable is still hoisted, but it won't be initialized until it's actually needed.

  var thing2; // this is uninitialized, initialize-undefined's cheaper cousin

if (!thing1) {

thing2 = 'abc';

For more info, check out this more casual explanation or go straight to the sauce.

"const" examples

To make a complicated concept easy, let's just say const creates a permanent version of a value you want to reuse. There are way too many ins and outs to explain why that isn't the case, and you shouldn't assume that all devs will know them. The functionality gets even more complicated when using const to declare Arrays or Objects, so instead I'll leave a rule of thumb that doesn't require technical knowledge:

Is this a time you'd use the ALL_CAPS_UNDERSCORE naming convention to convey a constant? Cool, then use const.

const will enforce a constant value, but it also needs to be communicated in your code. ALL_CAPS_UNDERSCORE naming is still the most recognizable convention for this.

If your code was a letter, it should read like this:

Dear future developer,

Here is a value that must never ever change.

Also, here is a global Object that references inconstant tuba data. The only thing constant about this Object is how *in*constant it will be.

Now that you know about these two things, I'm going to go get the tuba's info. It will take x amount of time.

With love, Some dev from the past

p.s. did you notice how much easier these paragraphs are to read with line-breaks between them?

And in browserspeak it looks like this:

  const API_URL = '/api/product/tuba';

var tuba = {};

getTuba().then((results) => {
    tuba =;
    tuba.randomNumber = Math.random();

function getTuba() {
    var promise = fetch(API_URL);
    return promise;


Like let, browsers hoist an uninitialized var of your const. Using these new tools hasn't relieved us of our responsibility to understanding how hoisting works. I think if anything, grasping this concept is the most effective way to prevent unintended or over use of these new variables.

So rest easy, var fans. Your old friend is still in the ECMAScript spec, and it plays a more important role than ever: improving readability.

430 0 0