Reactive programming with Pivotal's reactor
1 like1,852 views
The document provides an overview of reactive programming, particularly focusing on reactive streams and the usage of Flux in managing asynchronous event-driven programming. It explains key concepts like cold and hot Flux, operations for manipulating streams, and the advantages over traditional Java streams. The content also includes code examples and methodologies for creating, transforming, and subscribing to reactive streams.
Reactive programming with Pivotal's reactor
- 3. CloJure Cross language support Java 3 my_observable .map(Name::firstName) .map(String::toLowerCase) .filter(x -> x.startsWith("bob")) .dis+nct() .toList()
- 6. • Everything is a stream of messages • Flux emits event messages (“Observable” in rx-java land) • Subscribers consume messages • Fluxes are immutable (opera+ons return new Fluxes) Implements Publisher Emits messages from: • Remote services • Data feeds • Mouse events… Subscriber Consumes pushed messages: • onNext() • onError() • onComplete() Basic Model 6 Message stream Examples: • Events • Query results • Computa+ons • Even errors! Flux Subscriber
- 7. onNext() onComplete() onError() (onError() and onComplete() are terminal opera+ons.) A Subscriber receives noFficaFons:! public interface Subscriber<T>! void onNext(T t); void onComplete(); ! void onError(Throwable t); void onSubscribe(); ! ! 7
- 8. …AIaching a Subscriber 8 flux.subscribe( v -> handleHappyPath(v), // happy path error -> handleError(error), // error handler // (optional) () -> handle(“Done") // completion // (optional) );
- 9. Transform and combine Flux streams Flux .filter(…) .subscribe(…) 9 • Or even Flux • Or Flux .map(…) .subscribe(…) Each transforma+on returns a new resultant Flux. etc., etc., etc. .filter(…) .map(…) .subscribe(…) .filter(…) .map(…) .doOnNext(System.out::println) .subscribe(…)
- 11. Combine Fluxes to produce new Fluxes Examples: merge – merges elements as they arrive. zip – combines elements in sequence. firstEmikng – returns the first stream to emit an element. 11 • Methods take one Flux as input and return another Flux.
- 13. Transforming Fluxes Methods take one stream as input and return another stream. • Flux.take(n) – Takes first n elements only. • Flux.skip(n) – Skips first n elements, then takes the rest • Flux.distinct() – Returns a new Flux with duplicates eliminated. • Flux.distinctUntilChanged() – Eliminates consecutive duplicates. • Flux.filter((x)->condition()) – Retains elements matching filter condition. • Flux.map(Some::mapper) – Replaces elements with other elements. • Flux.flatMap(Some::mapper) – Replaces elements with Fluxes. 13
- 15. Difference between Java Stream API and Reac0ve 15 Java Streams ReacFve Streams Pull based Push based Basically a way to iterate collec+ons declara+vely A@ach to real-+me feeds Generally synchronous data Real-+me, concurrency, flow control Streams can only be used once Reac+ve streams are highly reusable No control of +ming Control back-pressure strategies No composi+on of streams Advanced composi+on and transforma+on Finite amount of data Data sizes from zero to infinity
- 17. Asynchronous Streams Built in support for concurrent publishers and subscribers • Flux.observeOn(scheduler) – Specifies the thread for the Observer • Flux.subscribeOn(scheduler) – Specifies the thread for the Subscriber Types of schedulers • Schedulers.immediate() – Parks current process and uses current thread • Schedulers.computa2on() – The system-assigned computa+on thread • Schedulers.io() – The system-assigned IO thread • Schedulers.trampoline() – Uses the current thread, once it is done here • Schedulers.newThread() – Uses a new thread • Schedulers.from(Executor) – On the named executor 17
- 18. Cold Fluxes and Hot Fluxes 18 • Cold Fluxes • Won’t begin pumping un+l a subscriber is a@ached. • Each subscriber receives all of the events, beginning from the historical first. • Hot Fluxes • Generally read live data, for example data feeds or mouse movements. • Begin pumping on connec+on. • Each subscriber gets the latest feeds as they pump
- 19. Addi0onal Cold Fluxes (for Development) • Flux.empty() – Completes on the first subscrip+on, without emikng any values. • Flux.never() – Emits no values and never completes. • Flux.error() – Emits an onError() no+fica+on immediately on every subscriber. No other values are emi@ed. • Flux.doOnNext() – Diagnos+cs 19
- 20. Crea0ng Hot Fluxes (Flowables) 20 o Call “publish” on a cold Flux ConnectableFlux<Long> hotFlux = coldFlux.publish(); o Call “connect” to start pumping, with or without subscribers hotFlux.connect(); hotFlux.subscribe( val -> System.out.println("Subscriber >> " +val));
- 21. Introducing Reac0ve Programming Pivotal’s Reactor Victor Grazi GMIT Core IT Demo
- 22. Victor Grazi GMIT Core IT Introducing Reac0ve Programming Pivotal’s Reactor Victor Grazi GMIT Core IT ? ? ? Q&A ?
- 23. Coun0ng le[ers 23 List<String> words = Arrays.asList( "the", "quick", "brown", "fox", "jumps", "over", "the", "lazy", "dog" ); Flux<Integer> lines = Flux.range(1, Integer.MAX_VALUE); Flux<String> wordsFlux = Flux.fromIterable(words); wordsFlux .flatMap(word -> Flux.fromArray(word.split(""))) .distinct() .sort() .zipWith(lines, (word, line) -> line + " " + word) .subscribe(System.out::println);
- 24. Metronome example 24 Flux<Long> fast = Flux.interval(Duration.ofSeconds(1)); Flux<Long> slow = Flux.interval(Duration.ofSeconds(3)); Flux<Long> clock = Flux.merge( fast.filter(t -> isFastTime()), slow.filter(t -> isSlowTime()) ); Flux<LocalDateTime> dateEmitter = Flux.interval(Duration.ofSeconds(1)) .map(t -> LocalDateTime.now()); Flux<LocalDateTime> localDateTimeFlux = clock.withLatestFrom(dateEmitter, (tick, date) -> date); localDateTimeFlux.subscribe(t -> System.out.println(t.format(DateTimeFormatter.ofPattern("yyyy-MM-dd hh:mm:ss"))));
- 25. Emi[er sample 25 SomeFeed<PriceTick> feed = new SomeFeed<>(); Flux<Object> priceFlux = Flux.create(emitter -> { SomeListener l = new SomeListener() { @Override public void priceTick(PriceTick event) { emitter.next(event); } @Override public void error(Throwable throwable) { emitter.error(throwable); } }; feed.register(l); }, FluxSink.OverflowStrategy.LATEST); ConnectableFlux<Object> connectableFlux = priceFlux.publish(); connectableFlux.connect(); connectableFlux.subscribe(System.out::println);