Although RxJs has a large number of operators, in practice we end up using a relatively small number of them.
And right after the most familiar operators that are also available in arrays (like map, filter, etc.), probably the first operator that we come across that is not part of the Array API but still very frequently used is the RxJs switchMap operator.
Let's see how this operator works by going over its behavior in a couple of very common situations:
- short-lived streams like for example HTTP Requests, that only emit one value
- long-lived streams such as for example the ones returned by AngularFire, which is an Angular library that provides some services for interacting with the Firebase real-time database and authentication
In the end, we will be covering a very common use case of the operator (emitting values that combine the output of multiple Observables).
For a complete comparison between the switchMap, mergeMap, concatMap and exhaustMap operators, have a look at this post:
Simulating HTTP and Firebase RxJs Streams
To focus on the operator behavior, let's introduce here a couple of utility functions that will help us to simulate certain stream types that we usually come across in our day-to-day development:
- Cold, short-lived, emit-once streams like Angular HTTP streams
- Cold, long-lived, emit-many values streams like AngularFire streams
Simulating these 2 very common types of RxJs streams will help us to understand better their characteristics and behavior, besides the
switchMap operator itself, so let's have a look at them.
How do Angular HTTP Observables work?
Let's start by simulating the Observables that are returned by the Angular HTTP library, these are probably the first stream type that we will encounter while learning Angular.
These are just one particular type of Observables that have a certain number of properties, not necessarily shared with other types of Observables:
- the HTTP observables are cold (or not live), meaning that they will not start emitting values until we subscribe to them
- these Observables only emit a single value or an error, and then after that they complete, so they are not long-lived Observables
- in most cases, we don't have to remember to unsubscribe from these Observables, because they will complete after emission
Simulating HTTP requests
With these features in mind, here is a function for creating an Observable with this behavior:
In this function we have used the
of utility function with a single value, to create an "emit once and complete" stream, and the
delay operator to make it asynchronous, just like an HTTP request.
To make sure that this function is working correctly, let's try it out. Let's use it to simulate a couple of HTTP requests:
These observables are cold, meaning that they will not emit values if they are not subscribed to. So let's subscribe to both Observables and have a look at the output:
Quickly outputting observable values for test purposes
Notice that we are using an abbreviated way of printing the output to the console, for example, the
http1$ subscription is equivalent to the following code:
In this example we are creating inline a function using the arrow operator, that takes the value and logs it. So that is why we can replace that with a direct reference to the
This shorthand notation comes in handy when for example debugging RxJs. With this in place, let's have a look at the output of these two subscriptions:
1 http1$ completed 2 http2$ completed
We can see that this observable has the usual behavior of an Angular HTTP Observable: each observable will emit only one value (it could also have emitted an error), and then the observable completes.
Introducing the Switch Map Operator
Let's then try the
switchMap operator to combine two HTTP requests, and see the result.
Let's simulate a request that saves some user data, and then reloads some other data that is impacted by that server-side modification.
This is how we could simulate this scenario:
Let's have a look at what the output of this program would look like:
simulating HTTP requests user saved data reloaded completed httpResult$
Given this output, let's break this down step-by-step to see what the switchMap operator is doing in this scenario:
saveUser$HTTP observable is said to be the source observable
- The output of the switchMap is the
resultObservable$constant, which we will refer to as the result observable
- if we don't subscribe to the result observable, nothing will happen
- if we subscribe to the result observable, that will trigger a subscription to the source Observable
- once the source Observable emits, the source value emitted is then passed on to the function that we have passed to the
- that function needs to return an Observable, that might be built using the source value or not
- that returned observable is said to be the inner observable
- the inner observable is then subscribed to, and its output is then emitted also by the result observable
- when the source observable completes, the result observable also completes
So as we can see the
switchMap operator is a great way of doing one HTTP request using the output of an initial request here, so this is one common way that we can use it.
But at this point, we might have a couple of questions in mind, like for example:
Why is this operator called switchMap?
Understanding why a functionality has a given name usually helps a lot in its understanding. So based on the name of the operator, we might ask the following questions:
- why the term
switch, concretely what is being switched, both from and to?
- why the term
map, what is being mapped in this example?
At this point, we might think the reason for the term
switch is that we are switching from the source observable to the inner observable that is being created by the function passed on the
switchMap, but as we will see that won't be the case.
The reason for the term
map is more or less clear, what is being mapped is the emitted source value, that is getting mapped to an observable using the mapping function passed to
So with this in mind let's now find out what is being switched.
Why we need a different type of stream to better understand switchMap
The problem with the type of short-lived HTTP-like streams that we have used so far, is that although we can already use
switchMap for a lot of use cases, its not 100% clear at this point how the operator works.
This is because the stream only emits once and then they complete, so its a very particular case. To better understand the
switchMap operator, we need to introduce a different type of streams: long-lived AngularFire-like streams.
Simulating AngularFire streams
Let's then introduce a utility function for simulating this new type of long-lived streams:
Using this utility function, let's have a look at this small program that simulates two Firebase-like streams:
How do the Firebase long-lived streams work?
Let's understand how this type of streams work, by looking at the output of the small program that we just wrote:
1: FB-2 0 2: FB-2 1 3: FB-2 2 4: FB-2 3 5: FB-1 0 6: FB-2 4 7: FB-2 5 8: FB-2 6 9: FB-2 7 10: FB-2 8 11: FB-1 1 12: FB-2 9 ...
This type of streams is quite different from HTTP streams. To understand how they work, let's break this down line by line, starting at the last one:
- maybe the most noticeable thing here is that both streams never complete, they keep emitting values!
- this type of streams will continue to emit new values (if a new value is available) until we unsubscribe from them
firebase2$observable emitted a few values, because it has an interval of only 1 second
firebase1$stream emitted fewer values because it has an interval of 5 seconds
Understanding the switchMap Operator
So with this in place, let's take these two long-lived streams and switch map one into the other, and see what happens - this new scenario will make more apparent what the operator is doing under the hood.
For example, let's try to guess the output of this small program:
Here is the program output, with some indentation added to it:
source value FB-1 0 inner observable 0 inner observable 1 inner observable 2 inner observable 3 source value FB-1 1 inner observable 0 inner observable 1 inner observable 2 inner observable 3 source value FB-1 2 inner observable 0 inner observable 1 inner observable 2 ...
The reason for the term Switch
Let's now learn why the term
switch is used to name the
switchMap operator, by breaking down this output step-by-step:
- just like in the case of the HTTP example, nothing will happen until the result observable is subscribed to
- the source observable will then emit the first value
- the source value will be mapped into an inner observable, which is the output of the mapping function passed to
- that inner observable will be subscribed to
- the result observable will emit the values also emitted by the inner observable
- we can see that the inner observable is emitting the values 0, 1, 2 that are the output of newly created interval
Why the term Switch?
But then something different than the HTTP example happens: because the source observable is long-lived, it will eventually emit a new value, in this case,
And from that new value onwards, the following happens:
- it looks like the long-running inner observable that was already at index 3 is no longer being used
- the mapping function is called again, and a new inner observable is created
- the new inner observable is now subscribed to
- the newly created inner observable will start emitting values again from index 0
- the result observable now is emitting the values emitted by the newly created inner observable, instead of the previously running inner observable
So what happened to the previously running inner observable?
That previous inner observable has been unsubscribed from, and so those values are no longer being used.
The result observable has
switchedfrom emitting the values of the first inner observable, to emitting the values of the newly created inner observable
And this explains the use of the term
switch in the name
The switchMap Operator in a Nutshell
Let's then summarize what we have learned so far: The
switchMap operator will create a derived observable (called inner observable) from a source observable and emit those values.
When the source emits a new value, it will create a new inner observable and
switch to those values instead. What gets unsubscribed from are the inner observables that get created on the fly, and not the source observable.
I hope this helps explain the operator, its name and some frequently used use cases for which you might want to use it (let me know in the comments).
But there is also a less well-known use case of this operator, that might come in very handy.
How to combine the output of multiple Observables?
Sometimes, we don't want to completely replace the values emitted by the source observable by the values emitted by the mapped inner obervable.
Instead of that, sometimes we want to combine the two outputs, and make them available both as part of the output value of the result observable.
The simplest way to do this is to apply the map operator to the inner observable, and return a tuple containing both values:
Inspecting the combined output of the two observables
Let's then see the end result on the console. As we can see, the value emitted by the result observable using the new selector function would be:
- the first element of the array is the course object, which is the output of the initial source observable HTTP call
- the second element of the array is an array of lessons, which is the value emitted by the inner observable
switchMap its a very useful and commonly used operator that is important to have a good grasp on, because we will be using it to implement many use cases.
This is one of those few operators that we will probably use in just about any application that we build.
I hope that these examples helped with this operator. To get notified when more posts like this come out on other commonly used RxJs operators, I invite you to subscribe to our newsletter:
If you would like to learn a lot more about RxJs, we recommend checking the RxJs In Practice Course course, where lots of useful patterns and operators are covered in much more detail.
If you are just getting started learning Angular, have a look at the Angular for Beginners Course:
Other posts on Angular
Have also a look also at other popular posts that you might find interesting:
- Getting Started With Angular - Development Environment Best Practices With Yarn, the Angular CLI, Setup an IDE
- Why a Single Page Application, What are the Benefits ? What is a SPA ?
- Angular Smart Components vs Presentation Components: What's the Difference, When to Use Each and Why?
- Angular Router - How To Build a Navigation Menu with Bootstrap 4 and Nested Routes
- Angular Router - Extended Guided Tour, Avoid Common Pitfalls
- Angular Components - The Fundamentals
- How to build Angular apps using Observable Data Services - Pitfalls to avoid
- Introduction to Angular Forms - Template Driven vs Model Driven
- Angular ngFor - Learn all Features including trackBy, why is it not only for Arrays ?
- Angular Universal In Practice - How to build SEO Friendly Single Page Apps with Angular
- How does Angular Change Detection Really Work ?