This is my take on Redux, the state container for JavaScript apps. It's oriented toward developers who like building React-style apps and Kotlin!
Why the name Bansa? Because it means "Nation" in Filipino. And a nations are "state containers." Get it!? Oh ho ho ho. Continue on for more, dear reader.
// First, add JitPack to your repositories
repositories {
...
maven { url "https://jitpack.io" }
}
// Base package. Everything else is built upon it!
compile 'com.github.brianegan.bansa:bansa:1.0.0-beta'
// If you'd like a more functional interface in Kotlin, go ahead and use the Kotlin Extensions.
compile 'com.github.brianegan.bansa:bansaKotlin:1.0.0-beta'
// Time Travel Dev Tools! Highly recommended you only use them
// for development, as the store takes up more memory. Note: You can use
// the provided UI, or write your own.
debugCompile 'com.github.brianegan.bansa:bansaDevTools:1.0.0-beta'
debugCompile 'com.github.brianegan.bansa:bansaDevToolsUi:1.0.0-beta'
"State Container" is pretty vague. So let me explain what I'm trying to accomplish with this little project: An easier way to write Android UIs & Apps. Perhaps an easy way to start would be a concrete analogy.
Think about List and RecyclerViews on Android. At a high level, you just give them some data and tell them how to render that data. When the data updates, you call notifyDataSetChanged
to inform the View and it redraws everything for ya. Nice and simple. And heck, with RecyclerView, it'll even perform some pretty impressive animations with just a couple lines of code!
That's what I'm going for with this project: I want a simple way to declare the way my interface should look, and when the data changes, everything should re-render for me!
So where does Bansa fit into that picture, one might ask? It doesn't say anything on the box about making my life easier as a UI developer!?
Bansa doesn't handle the UI magic. That's left to other tools, namely Anvil. Bansa is responsible for holding, or "containing," the state of your application, and informing the UI layer (or database layer, or logging layer, etc) about updates to the state of the application.
The examples in this project simply show one person's vision for how we could think about Android App development, and whether we could make it easier and more fun.
We'll demonstrate using a simple counter example!
All we need for a simple counter example is the value of one counter. In one delicate line with Kotlin:
data class ApplicationState(val counter: Int = 0)
Actions are payloads of information that send data from your application to your state container, which we'll call the store
from here on out. They are used to determine what actions your store
should respond to. You send them from your application to a store
with the store.dispatch(ACTION)
method.
So here are the three actions we need for our counter app:
object INIT : Action
object INCREMENT : Action
object DECREMENT : Action
Actions describe the fact that something happened, but don’t specify how the application’s state changes in response. This is the job of a reducer.
Let's see some code and we'll chat about it afterwards:
val reducer = Reducer<ApplicationState> { state, action ->
when (action) {
is INIT -> ApplicationState()
is INCREMENT -> state.copy(counter = state.counter.plus(1))
is DECREMENT -> state.copy(counter = state.counter.minus(1))
else -> state
}
}
So what's happening here? A reducer is a function that takes two arguments: the current state of the application, and the action that was fired. It returns an updated version of the state.
In this example, when the INIT
action is fired, we want to initialize the state of our application. Therefore, we return a new instance.
When the INCREMENT
action is fired, we want to simply increase the counter by 1.
If DECREMENT
is fired, we'll need to decrease the counter by 1.
And that's all there is to it: We're simply describing how the state should change in response to an action.
Now that we've gotten everything setup, we can create our state container!
val counterStore = BaseStore(ApplicationState(), reducer);
And that's it! Now you've got a store. Woohoo! So what now?
The following code will dispatch an action that travels first through any middleware (we'll chat about those later), then through the reducers, generally producing a state change.
counterStore.dispatch(INCREMENT)
Say you want to know what the current state of the app is. That's simple:
counterStore.getState() // After the INCREMENT, state is now (counter = 1)
counterStore.subscribe({
textView.setText(store.getState())
})
ZOMG MY TEXT JUST UPDATED DUE TO A STATE CHANGE!!!
I know what you're saying: "Bri, I've seen all this before. It's called an EVENT BUS." And you know what? You're pretty much right. This isn't anything radical, it's just gluing some pieces together in a different way. But just think about how we can use this pattern to simplify our UIs.
Ok, so now we have to chat about Anvil. It's a simple way to write UIs in Java & Kotlin, and it might be a bit different than what you're used to. With Anvil, you simply describe your UI in Java / Kotlin code (not XML -- give it a few minutes, I think you'll fall in love), update the state, and call Anvil.render()
. Then everything just updates! We've done it! We've achieved the goal set out in the opening paragraphs!!!
So here's an example view. It's job is to create a linearLayout with three child views: A text view to display the current count, an plus button, and a minus button.
When someone clicks on the plus button, we want to dispatch an INCREMENT
action. When someone clicks the minus button, we want to dispatch a DECREMENT
action.
linearLayout {
size(FILL, WRAP)
orientation(LinearLayout.VERTICAL)
textView {
text("Counts: ${store.state.counter.toString()}") // The counter from our state model!
}
button {
size(FILL, WRAP)
padding(dip(10))
text("+")
onClick(store.dispatch(INCREMENT))
}
button {
size(FILL, WRAP)
padding(dip(5))
text("-")
onClick(store.dispatch(DECREMENT))
}
}
And now, we hook anvil and Bansa up together:
counterStore.subscribe({
Anvil.render()
})
That's right: When a user clicks "+", the increment action will be fired, the reducer will update the state, and our UI will auto-render with the new info. WHAAAAAAAAAAT.
Maybe one of the cooler bonuses about this pattern is that it happens to lend itself well to time-travel style Dev tools! What are time travel dev tools you say? What if, for every action you take in an application, you could replay what happened? Then you could step forward and backward through that series of actions to see the steps along the way. You can totally do that with Bansa!
More documentation needs to be written about implementing the Time Travel Dev tools, but for now, please see the Counter example as a reference implementation! It should be possible to get the Dev Tools up and running with only a small amount of code.
There are a progression of examples that can be found in the "Examples" folder. If you're interested in progressing through them, the suggested order is:
- Counter
- Counter pair
- List of counters
- List of counters variant
- Random gif
- List of trending gifs
- Todo List
Write sections for:
- Async Actions
- Middleware
- Combining reducers
- Breaking down apps into smaller parts
- How views should asynchronously request / remove data from the store upon instantiation and disposal if it's safe.
- What is the role of the database? Store actions? Store models?
- Could views describe their data requirements as queries, similar to Falcor or GraphQL?
If you're thinking of writing your own version of Redux in Java: I'd say, go for it! This libray has benefitted tremendously from the various versions that have been written thus far, and the current version is an attempt to synthesize the best parts of each implementation. When evaluating the various libs, I came up with a bucket list for what this library should accomplish for a 1.0 release:
- Idiomatic Java and Kotlin usage.
- Low method count
- Easy to pick up, even if you come from a less functional background.
- Support for all basic Redux concepts in the smallest possible package. This includes the Store, Middleware, Reducers, and Subscribers/Subscriptions, and combining Reducers.
- Dev Tools that allow for Time Travel debugging
- Interface that allows for sharable middleware & reducers (such as Logging middelware, or Undo/Redo reducer helpers)
- Well tested
If you like Buzzwords, then boy howdy, have you found yourself the right repo.
Bansa itself has one important dependency, RxJava, and is written in Kotlin.
The examples use:
- Kotlin (yay! it's so lovely.)
- Anvil for UI
- RxJava for Observables
- OkHttp for, uh, http
- Moshi for Json parsing
- Picasso for image loading
- JUnit for Unit Testing
- AssertJ to make dem assertions all fluent like
I'd also like to add that this library is a complete , these ideas aren't new in any way. Credit has to be given to the following projects, listed in autobiographical order.
- React
- Om
- ClojureScript
- Elm
- Redux
- CycleJs
- redux-java - by my coworker and friend, Guillaume Lung
- re-frame