A webinar recently inspired me that egghead.io hosted with Andre Staltz and I wanted to share what I learned. Before the webinar, I was unfamiliar with RxJS and it was the first time I was exposed to the observer pattern. Before it was broken down, Observers seemed like magic.
JavaScript has multiple APIs that use callback functions that all do nearly the same thing with slight variations.
Streams
stream.on('data', data => {
console.log(data)
})
stream.on('end', () => {
console.log("Finished")
})
stream.on('error', err => {
console.error(err)
})
Promises
somePromise()
.then(data => console.log(data))
.catch(err => console.error(err))
Event Listeners
document.addEventListener('click', event => {
console.log(event.clientX)
})
The rough pattern you see is that there is an object, and inside the object, you have some method that takes a function, in other words, a callback. They're all solving the same problem, but in different ways, this causes you to have to carry the mental overhead of remembering the specific syntax for each of these APIs. That's where RxJS comes in. RxJS unifies all of this under one common abstraction.
So what even is an observable? It's an abstraction in the same way that arrays, functions, or objects are all abstractions. A promise can either resolve or reject, giving you back one value. An observable is capable of emitting values over time. You could consume streams of data from a server or listen for DOM events.
π Observable Skeleton
const observable = {
subscribe: observer => {
},
pipe: operator => {
},
}
Observables are just objects that contain a subscribe
and pipe
method. Wait, what's going on here? What's an observer, or an operator? Observers are just objects that contain the callback methods for next
, error
, and complete
. The subscribe
method consumes an observer and passes values to it. So observable is acting as a producer, and the observer is its consumer.
π An Observer
const observer = {
next: x => {
console.log(x)
},
error: err => {
console.log(err)
},
complete: () => {
console.log("done")
}
}
Inside of that subscribe
method you pass some form of data to the observer's methods.
Subscribe Method
const observable = {
subscribe: observer => {
document.addEventListener("click", event => {
observer.next(event.clientX)
})
},
pipe: operator => {
},
}
Here we are just listening for clicks made anywhere in the document. If we ran this and made a call to observable.subscribe(observer)
, we would see the x coordinates of your clicks showing up in the console. So what about this pipe
method? The pipe
method consumes an operator and returns a function, and makes a call to the resulting function with the observable.
Pipe Method
const observable = {
subscribe: observer => {
document.addEventListener("click", event => {
observer.next(event.clientX)
})
},
pipe: operator => {
return operator(this)
},
}
Cool but what's an operator? Operators are for transforming your data. Arrays have operators, like map
. map
lets you take a step back and run some function over everything in the array. You could have an array and then another array that is a mapped version of the first.
Let's write a map
function for our observable.
πΊοΈ Map Operator
const map = f => {
return observable => {
subscribe: observer => {
observable.subscribe({
next: x => {
observer.next(f(x))
},
error: err => {
console.error(err)
},
complete: () => {
console.log("finished")
}
})
},
pipe: operator => {
return operator(this)
},
}
}
A lot is going on here so let's break it down.
const map = f => {
return observable => {
Here we are passing in a function and returning a function that expects an observable. Remember our pipe
method?
pipe: operator => {
return operator(this)
},
To run the operator on the observable, it needs to get passed into pipe
. pipe
is going to pass the observable it's called on into the function that our operator returns.
subscribe: observer => {
observable.subscribe({
Next, we are defining the subscribe
method for the observable that we are returning. It expects an observer, which it receives in the future when .subscribe
gets called on the returned observable, either through another operator or explicitly. Then, a call gets made to observable.subscribe
with an observer object.
{
next: x => {
observer.next(f(x))
},
error: err => {
console.error(err)
},
complete: () => {
console.log("finished")
}
}
In the observer's next
method you can see that a call to a future observer's next
is made with the function that we originally passed into map
and an x
value passed into next
. Let's run our new map
operator on our observable!
observable
.pipe(map(e => e.clientX))
.pipe(map(x => x - 1000))
.subscribe(observer)
That final subscribe
is needed or none of the operations inside of those operators execute, that's because they are all wrapped up in their observer's subscribe
methods. In those subscribe
methods is a call to subscribe
the previous observer in the chain, but the chain has to begin somewhere.
So let's follow what happens when this runs.
- The first pipe gets called on observable,
map
gets curried withthis
-
map
is called withe => e.clientX
and it returns a function - The function gets called with the original
observable
and an observable gets returned- We'll call it observable2
-
pipe
is called onobservable2
and curriesmap
withthis
-
map
is called withx => x - 1000
and it returns a function - That function gets called with
observable2
and an observable gets returned- We'll call it observable3
-
.subscribe
gets called onobservable3
with an observer passed in -
.subscribe
gets called onobservable2
with the operator's observer passed in -
.subscribe
is called on the original observable with the operator's observer passed in - A click event happens with a
clientX
of100
-
observer2.next(100)
gets called -
observer3.next(100)
gets called -
observer.next(-900)
gets called and logs-900
to the console. - Done!
You can see the chain happen here. When you call subscribe
you are asking for information, each link asks the previous link in the chain for it until it reaches the data and the next
method from its observer gets called. That data then rises back up the chain, getting transformed along the way, until it then reaches the final observer.
Here is the code in its entirety.
const observable = {
subscribe: observer => {
document.addEventListener("click", event => {
observer.next(event.clientX)
})
},
pipe: operator => {
return operator(this)
}
}
const observer = {
next: x => {
console.log(x)
},
error: err => {
console.log(err)
},
complete: () => {
console.log("done")
}
}
const map = f => {
return observable => {
subscribe: observer => {
observable.subscribe({
next: x => {
observer.next(f(x))
},
error: err => {
console.error(err)
},
complete: () => {
console.log("finished")
}
})
},
pipe: operator => {
return operator(this)
},
}
}
Top comments (8)
I didn't find the list of steps useful, but having code for
observable
,observer
,operator
(map
) and the use case fitting on a page allowed me to finally grok howRXJS
works on a fundamental level.Choosing
map
as the operator really resonated with me!Thanks for sharing π
Thanks for the article. It is well written.
One comment: Arrow function may not work when it uses 'this'.
Changing it to
function
worksAnother fix in the line to pass the event instead of event.clientX
Good catches! I forgot about the scope differences between
=>
andfunction
Congrats for the article.
U explain in a easy way what's producer and consumer, I never had think like ur example haha
But about explain a map operator (or any operator) I really think is massive to rxjs begginers
This is great! Really appreciate the breakdown π
Good job, fixed few things, here is the working code:
const observer = {
next: x => {
console.log(x)
},
error: err => {
console.log(err)
},
complete: () => {
console.log("done")
}
}
const observable = {
subscribe: observer => {
document.addEventListener("click", event => {
observer.next(event.clientX)
})
},
pipe: function (operator) {
return operator(this)
}
}
mostly here, correct pattern is return and return! ->
const map = function (f) {
return observable => {
return {
subscribe: observer => {
observable.subscribe({
next: x => {
observer.next(f(x))
},
error: err => {
console.error(err)
},
complete: () => {
console.log("finished")
}
})
},
pipe: function (operator) {
return operator(this)
},
}
}
}
//observable.subscribe(observer);
observable
.pipe(map(e => e.clientX))
.pipe(map(x => x - 1000))
.subscribe(observer)
aworking example for node.js using std-input is here too:
github.com/babak2000ir/rxjs-simula...
Best explanation Iβve read on the web. Thanks! It would be awesome if you had another example of a slightly more complicated operator than map also.
yep. Still have a lot to learn.
I thought I followed up until you said you watched a webinar and then I lost you. That's on me tho not your article.