Architectures with Kotlin Multiplatform

In the current wave of cross platform solutions, Kotlin’s multiplatform is shaping up to be a unique proponent of code reuse. Unlike write once run everywhere, Multiplatform offers a unique space of reusability and formalization of code across platforms and also removes dependency on custom VMs or runtime. Modern apps are build on top of abstractions and architectures which make sure code-bases remain mature, testable and error free. They on the other hand also introduce a lot of boilerplate code. This presents an opportunity of code reuse with help of Kotlin multiplatform. Through this talk, we’ll see how can harness power of Kotlin Multiplatform to create apps using architectures such as MVP, MVI etc and use Kotlin native to build an abstraction layer across platforms. We’ll also go through usage of various multiplatform libraries for database and network management.

Transcript

1. Architectures with Kotlin Multiplatform Jitin Sharma GO-JEK @_jitinsharma
2. Most popular among Android Developers Was designed with platform agnostic behaviour. Has a lot of features which can targeted outside JVM environment.
3. JVM/Server Android .kts Javascript Native
4. Kotlin Native Can target iOS, Windows, MacOS, Linux etc No Runtime environment
5. Interop C/Objective C interop provided System frameworks are automatically imported in case of iOS. Types are mapped with Swift/Objective C.
6. Multiplatform One Language to rule them all Started with Kotlin 1.2 Still in experimental phase. Built in tooling with IntellijIDEA.
7. “Multiplatform is not cross platform” –Kotlin
8. common android iOS js server project - Presentation logic - ui-logic - Data models - Helper/utilities - Networking - Storage - Data models - Networking? - Storage?
9. onStart onResume … viewDidLoad viewDidAppear … Async networking Async storage Kotlin Swift
10. app commonMain main iosMain iosApp .kt,xml .kt .kt Imports common libraries Imports iOS libraries Builds Android app Builds .framework .swift, xib
12. Common Logic dependencies { implementation kotlin('stdlib-common') }
13. val names = mutableListOf("Bumblebee", "Optimus Prime", "Megatron") names.sort() Foo.kt
14. expect fun <T : Comparable<T>> MutableList<T>.sort(): Unit val names = mutableListOf("Bumblebee", "Optimus Prime", "Megatron") names.sort() public actual fun <T : Comparable<T>> MutableList<T>.sort(): Unit { if (size > 1) java.util.Collections.sort(this) } MutableCollectionsJVM.kt CollectionsH.kt Foo.kt
15. expect & actual expect fun … expect val… expect class… Actual Platform declarations
16. Client Server
17. NetworkClient.kt class NetworkClient { private val client = HttpClient() fun get(url: String, callback: (String) -> Unit) { GlobalScope.launch(ApplicationDispatcher) { val result: String = client.get { url(url) } callback(result) } } } internal expect val ApplicationDispatcher: CoroutineDispatcher
18. internal expect val ApplicationDispatcher: CoroutineDispatcher internal actual val ApplicationDispatcher: CoroutineDispatcher = Dispatchers.Default AndroidDispatcher.kt
19. internal expect val ApplicationDispatcher: CoroutineDispatcher import platform.darwin.dispatch_async import platform.darwin.dispatch_get_main_queue import platform.darwin.dispatch_queue_t internal actual val ApplicationDispatcher: CoroutineDispatcher = NsQueueDispatcher(dispatch_get_main_queue() internal class NsQueueDispatcher(private val dispatchQueue: dispatch_queue_t) : CoroutineDispatcher() { override fun dispatch(context: CoroutineContext, block: Runnable) { dispatch_async(dispatchQueue) { block.run() } } } IOSDispatcher.kt
20. Serialization @Serializable data class AllData( val avatar_url: String?, val bio: String?, val blog: String, .... ) import kotlinx.serialization.*
21. Persistence - SQLDelight CREATE TABLE Photos ( id INTEGER NOT NULL, title TEXT NOT NULL ); insert: INSERT INTO Photos (id, title) VALUES (?, ?); getAll: SELECT * FROM Photos;
22. Persistence expect val db: Database db.photoQueries.getAll() -> List<Photo> //iOS actual val db = Database(NativeSqliteDriver(Database.Schema, "photo.db")) //Android actual val db = Database(AndroidSqliteDriver(Database.Schema, context, “photo.db"))
25. interface MainView { fun displayData(data: DisplayData) fun showLoader() fun hideLoader() }
26. class DataRepository { private val api = NetworkApi("https://github.com") fun getData(username: String, callback : (Data) -> Unit) { val data = api.getAll(username) { callback(it) } } }
27. class MainPresenter(private val view: MainView, private val repository: DataRepository) { fun loadData(userName: String) { if (userName.isNullOrEmpty()) { view.showError(USER_NAME_NOT_VALID) } else { view.showLoader() repository.getData(userName) { showData() view.hideLoader() } } } }
28. extension ViewController: MainView { func showLoader() { activityIndicator.startAnimating() } func hideLoader() { activityIndicator.stopAnimating() } func displayData(data: DisplayData) { .... } } ViewController.swift
29. PhotoListState
30. PhotoListState Store States Reducers Middlewares Action
31. data class Photo( @SerializedName("id") val id: String, @SerializedName("title") val title: String, ...... ) gson serialization
32. expect val store: Store<AppState> expect class AppState : StateType expect fun appReducer(action: Action, state: AppState?): AppState class LoadRecentPhotos : Action class SerializeResponse(val response: String) : Action Actions.kt Common
33. actual val store: Store<AppState> = Store( state = AppState(), reducer = ::appReducer, middleware = listOf(networkMiddleWare, databaseMiddleWare) ) actual data class AppState(val photoListState: PhotoListState? = null) : StateType actual fun appReducer(action: Action, state: AppState?): AppState { return AppState(photoListState = photoListReducer(action, state?.photoListState)) } Android
34. class MainActivity : AppCompatActivity(), StoreSubscriber<PhotoListState?> { override fun newState(state: PhotoListState?) { state.shouldUpdate { val photos = it.photos setImageAdapter(photos) } } }
35. Performance Generated framework binaries are smaller in size. Few extra classes are generated for various kotlin types For Android/JVM there is not much impact.
36. Debugging For Android/JVM projects, both common code and android code can be debugged right from IDE. For iOS, generated framework files can be debugged accordingly.
37. Testing Shared code test cases can be written out of IntelliJ using kotlin.test Platform specific execution using platform test libraries. Existing platform libraries(espresso, xctest) can still be used.
38. DIY Multiplatform Libraries common android iOS ./gradlew build library.jar library.framework Multiplatform Publishing library-common.jar library-android.jar library-ios.klib .module
39. DIY Multiplatform Libraries common android iOS ./gradlew build library.jar library.framework Multiplatform Publishing library-common.jar library-android.jar library-ios.klib .module maven, pod
40. What’s NOT great Dependence of Multiplatform libs for better code sharing Still experimental
41. What’s great Language - Kotlin No external dependencies as in custom runtime/ VM Pragmatic APIs Platform specific libs with common Kotlin code
42. Multiplatform Libraries https://github.com/ktorio/ktor https://github.com/Kotlin/kotlinx.coroutines https://github.com/square/sqldelight/ https://github.com/russhwolf/multiplatform-settings https://github.com/ReKotlin/rekotlin-multiplatform
43. Thanks @_jitinsharma