JSFeeds: swizec.com - JavaScript promises are just like monads and I can explain both in less than 2 minutes

Thursday, 21 September, 2017 UTC

JavaScript promises are just like monads and I can explain both in less than 2 minutes

Summary

There’s a joke in functional programming that once you understand monads, it becomes impossible to explain them to others. The amount of monad tutorials on the internet is growing almost exponentially

Monads are a funky concept that’s nearly impossible to understand in all of its nuance. Maybe I’m just not smart enough. That’s why I’m not going to explain any of that.

Screw the mathematical definition. Look at this shit

o.O wat O.o

Here’s a handwavy explanation instead: Monads are like a bubble. They wrap your dirty values and protect the rest of your code from weird effects.

I used the continuation monad as an example because it is similar to JavaScript Promises. A way to talk about the future. 1

Let’s pretend you’re a mouse looking for the ultimate question to life, the universe, and everything. You build a super computer that will calculate the answer, and you call it Earth.

You know Earth will take around 4.54 billion years to calculate the question. But you’re writing the code right now. You can’t wait 4.54 billion years to finish your project.

What do you do? You put Earth in a time bubble.

Like this:

function getQuestion() {
    return Earth()
       .then(ArthurDent => ArthurDent.subconscious())
       .then(subconscious => Scrabble.output(subconscious))
}

In The HitchHiker’s Guide to The Galaxy, Arthur Dent was the final result of a 4.54 billion year calculation. He mindlessly picked letters out of a pile of Scrabble™ tiles, and the result produced the ultimate question.

That means our getQuestion function first constructs an Earth, then gets ArthurDent, accesses his subconsciousness, and uses Scrabble™ to print the result. 🤙

Here’s what that looks like in practice:

Let’s compare promises to callbacks. You’re likely to have met with callbacks before, JavaScript is full of them.

let theFuture = function (callback) {
    setTimeout(callback, 5000);
}

theFuture(() => {
    console.log("It is now 5 seconds later");
});

We create a function called theFuture. It waits 5 seconds, then calls our callback. Inside the callback, we print something.

Using a promise, that same code looks like this

let theFuture = new Promise((resolve, reject) => {
    setTimeout(resolve, 5000);
})

theFuture.then(() => {
    console.log("It is now 5 seconds later");
});

We turn theFuture into a Promise and trigger its resolve method after 5 seconds. The Promise wrapped everything inside its body into a protective time bubble.

With a promise, we can pass theFuture around and do all sorts of stuff. But if we want to access the future, we have to use .then and give it a function.

When I first saw this, I thought “So what’s the big deal? This is the same as callbacks.”

So here’s the big deal: Once you’re in a promise, you’re always in a promise. Because promises are like monads

Check out this simplified real world example from my day job.

makePurchase(){
    this.paymentView
          .fetchPaymentInfo()
          .then(paymentInfo  => this.finishPurchase(paymentInfo))
      .then(token => this.showSuccessModal())
      .catch(error => console.error(error));
}

fetchPaymentInfo(){
    if (this.paypal) {
        return this.getPaypalPaymentInfo(); // returns promise
    }else{
        return this.getCCPaymentInfo(); // returns promise
    }
}

getPaypalPaymentInfo(){
    return this.brainTreeClient
                         .tokenizePaypal()
                         .then(response => {
                 return response.paymentInfo; 
                         });
}

finishPurchase( paymentInfo ) {
    return fetch('/purchase/path')
                        .then(response => response.json())
                        .then(json => {
                            if (json.token) {
                                return json.token;
                            }else{
                                throw new Error("Purchase failed")
                            }
                        });
}

Don’t worry about the behind-the-scenes details of that code. Here’s what you should focus on:

The main makePurchase function does everything through Promise access. The .then and .catch methods are like peeking into the Promise time bubble. You’re saying: Once this time bubble resolves, I want to do so and so with the result.
The fetchPaymentInfo is the first method in our chain that creates a Promise. From then onwards, we can access returned values only through .then and .catch. It uses getPaypalPaymentInfo and getCCPaymentInfo to talk to Braintree, which is an operation that takes some time.
braintreeClient.tokenizePaypal() returns a promise. There’s no need to wrap this in another Promise inside fetchPaymentMethod. You can return it like any normal value.
getPaypalPaymentInfo uses .then to look into the Braintree response. Since you’re in a Promise, you can return flat values without worry. They’re already wrapped in a Promise.
As a result, we can chain multiple .then calls in makePurchase. Some methods return a regular value, some return a Promise. JavaScript don’t care, it’s all the same because everything returned from a Promise is a Promise.

And that’s why Promises are just like monads.

Oh, and error handling. Don’t worry about that either. As long as there’s a .catch call somewhere in the chain, you’re good. Errors bubble up through the chain of Promise look-into-s until they encounter a .catch.

Hope that helps, it took me months of practice to grok 🤓

Technically continuations are more similar to callbacks, but bear with me. ↩

The post JavaScript promises are just like monads and I can explain both in less than 2 minutes appeared first on A geek with a hat.

... more @ swizec.com

swizec.com