Mobius is a functional reactive framework for managing state evolution and side-effects, with add-ons for connecting to Android UIs and RxJava Observables. It emphasizes separation of concerns, testability, and isolating stateful parts of the code.
To learn more, see the wiki for a user guide. To see Mobius in action, check out the sample TODO app based on the app from Android Architecture Blueprints. You can also watch a talk from Android @Scale introducing Mobius.
Mobius is in Production status, meaning it is used in production in Spotify Android applications, and that we consider the APIs to be stable and the implementation bug-free. We will not make backwards-compatibility-breaking changes.
Mobius is currently built for Java 7 (because Java 8 is not fully supported on all versions of Android), hence the duplication of some concepts defined in java.util.function
(see com.spotify.mobius.functions
).
When using Mobius, we recommend using Kotlin or Java 8 or later, primarily because of the improved type inference and because using lambdas greatly improves readability and conciseness of code.
The latest version of Mobius is available through Maven Central (LATEST_RELEASE below is ):
implementation 'com.spotify.mobius:mobius-core:LATEST_RELEASE'
testImplementation 'com.spotify.mobius:mobius-test:LATEST_RELEASE'
implementation 'com.spotify.mobius:mobius-rx:LATEST_RELEASE' // only for RxJava 1 support
implementation 'com.spotify.mobius:mobius-rx2:LATEST_RELEASE' // only for RxJava 2 support
implementation 'com.spotify.mobius:mobius-android:LATEST_RELEASE' // only for Android support
implementation 'com.spotify.mobius:mobius-extras:LATEST_RELEASE' // utilities for common patterns
The goal of Mobius is to give you better control over your application state. You can think of your state as a snapshot of all the current values of the variables in your application. In Mobius, we encapsulate all of the state in a data-structure which we call the Model.
The Model can be represented by whatever type you like. In this example we'll be building a simple counter, so all of our state can be contained in an Integer
:
Mobius does not let you manipulate the state directly. In order to change the state, you have to send the framework messages saying what you want to do. We call these messages Events. In our case, we'll want to increment and decrement our counter. Let's use an enum
to define these cases:
enum CounterEvent {
INCREMENT,
DECREMENT,
}
Now that we have a Model and some Events, we'll need to give Mobius a set of rules which it can use to update the state on our behalf. We do this by giving the framework a function which will be sequentially called with every incoming Event and the most recent Model, in order to generate the next Model:
class CounterLogic {
static Integer update(Integer model, CounterEvent event) {
switch (event) {
case INCREMENT: return model + 1;
case DECREMENT: return model - 1;
}
}
}
With these building blocks, we can start to think about our applications as transitions between discrete states in response to events. But we believe there still one piece missing from the puzzle - namely the side-effects which are associated with moving between states. For instance, pressing a "refresh" button might put our application into a "loading" state, with the side-effect of also fetching the latest data from our backend.
In Mobius, we aptly call these side-effects Effects. In the case of our counter, let's say that when the user tries to decrement below 0, we play a sound effect instead. Let's create an enum
that represents all the possible effects (which in this case is only one):
enum CounterEffect {
PLAY_SOUND,
}
We'll now need to augment our update
function to also return a set of effects associated with certain state transitions. To do this we'll implement the Update
interface like so:
class CounterLogic implements Update<Integer, CounterEvent, CounterEffect> {
public Next<Integer, CounterEffect> update(Integer model, CounterEvent event) {
switch (event) {
case INCREMENT:
return next(model + 1);
case DECREMENT:
if (model == 0) {
Set<CounterEffect> soundEffect = effects(CounterEffect.PLAY_SOUND);
return dispatch(soundEffect);
} else {
return next(model - 1);
}
}
throw new IllegalStateException("Unhandled event: " + event);
}
}
Mobius sends each of the effects you return in any state transition to something called an Effect Handler. Let's make one of those now by implementing the Connectable
interface:
class CounterEffectHandler implements Connectable<CounterEffect, CounterEvent> {
public Connection<CounterEffect> connect(Consumer<CounterEvent> output) {
return new Connection<CounterEffect>() {
@Override
public void accept(CounterEffect effect) {
if (effect == CounterEffect.PLAY_SOUND) {
Toolkit.getDefaultToolkit().beep();
}
}
@Override
public void dispose() {}
};
}
}
Now that we have all the pieces in place, let's tie it all together:
public static void main(String[] args) {
// Let's make a Mobius Loop
MobiusLoop<Integer, CounterEvent, CounterEffect> loop = Mobius
.loop(new CounterLogic(), new CounterEffectHandler())
.startFrom(0);
// And start using our loop
loop.dispatchEvent(CounterEvent.INCREMENT); // Model is now 1
loop.dispatchEvent(CounterEvent.DECREMENT); // Model is now 0
loop.dispatchEvent(CounterEvent.DECREMENT); // Sound effect plays! Model is still 0
}
This covers the fundamentals of Mobius. To learn more, head on over to our wiki.
We're using Google's auto-formatter to format the code. The build pipeline is set up to fail builds that aren't correctly formatted. To ensure correct formatting, run
./gradlew format
This project adheres to the Open Code of Conduct. By participating, you are expected to honor this code.