Reactive Streams
- Provides standard for asynchronous stream, non-blocking applications.
- Use “Pull” model (Subscriber ask for data) instead of “Push” model (Publisher publish data)
There are 4 interfaces
Publisher<T> -> provides bounb/unbound elements that publishing according to Sucscriber(s)
- void subscribe(Subscriber<? super T> s); : calls to subscribe to publisher, each call will start new Subscription.
Processor<T, R> -> be both Publisher and Sucscriber
Subscriber<T> ->
- void onSubscribe(Subscription s); : invoke after calling Publisher.subscribe
- void onNext(T t); : data sent by Publisher as request(one at a time)
- void onError(Throwable t); to notifies errors and no more elements will be published
- void onComplete(); : to notifies all data sents and no more elements will be published
Subscription ->
- void request(long n); : request n elements from Publisher
- void cancel(); : inform Publisher to stop publishing data to this Subscription
Java 9 Flow interfaces
Later, Java team includes Reactive Streams specification in to JDK, so calls Java 9 Flow interface
Implementation of Reactive Streams
- RxJava
- Reactor
- Akka stream
RxJava
- RxAndroid is an extension to RxJava
- Observable =
Publisherimeplementation that emits 0..n elements - There are also others types Flowable, Single, Maybe
- Disposable is subscription that we can use for cancellation
Reactor
- Spring webflux is base of Reactor
- Cold stream = static, fix lenght stream of data. For example, harcoded Flux.just(…).
- Hot stream = dynamic, infinite stream of data. For example, data from Kafka.
- Mono =
Publisherimeplementation that emits 0..1 elements - Flux =
Publisherimeplementation that emits 0..n elements