RxJS and Reactive Programming - Modern Web UI - May 2015
Download as PPTX, PDF34 likes56,761 views
The document discusses the use of RxJS and reactive programming in frontend development, emphasizing the importance of observables for managing asynchronous data and event handling. It highlights the benefits of using observables for stream processing, such as reduced memory allocation and improved efficiency, especially in scenarios involving web sockets. Additionally, it provides insights into data flow management and the role of observables in controlling side effects within applications.
RxJS and Reactive Programming - Modern Web UI - May 2015
- 1. RxJS and Reactive Programming Ben Lesh Senior UI Engineer Netflix Edge Tools & Insights @benlesh
- 2. Observables
- 3. Observables • TC39 proposal to add them to ES2016 (ES7) • Angular 2 uses them and supports them in a first class way • ReactJS will use them and support them in a first class way • (Ask me about Ember after :P)
- 4. Front End Development Mostly synchronous or asynchronous?
- 5. We tend to think synchronously We write code in blocks that are read top to bottom, left to right. If this, then this, else this…
- 6. But What Are We Doing With Code? • Handling User Input • Animations • AJAX/JSONP • Web Sockets/SSE • Updating DOM
- 7. Most of these things are ASYNC! • Handling User Input • Animations • AJAX/JSONP • Web Sockets/SSE • Updating DOM
- 8. Async in JavaScript usually looks like this A function that gets called 0 to N times over time, generally being passed a value
- 9. Asynchronous values Can be represented as collections!
- 10. Collections today We have Iterables
- 11. Iterable
- 12. Observable is basically an Iterable turned inside-out
- 13. RxJS Observables
- 15. RxJS Observables (with error handling)
- 16. In fact, this should look familiar…
- 17. But Observables handle more than one value, so we need to know when they’re done
- 18. Observables Any number of values over any amount of time
- 19. SO WHAT? We can represent values over time as a collection… big deal.
- 20. Set operations! • Map • Reduce • Filter • Concat • Merge • Zip • FlatMap • Take • Skip
- 21. Plus async set operations! • Map • Reduce • Filter • Concat • Merge • Zip • FlatMap • Take • Skip • Buffer • Window • CombineLatest • Scan • FlatMapLatest • WithLatestFrom
- 22. Confused Yet? Code Example!
- 23. Observables for collections? I have arrays and iterables! Arrays already have map, filter, reduce
- 24. Demo day at Netflix!
- 25. We hooked it up to prod data and…
- 26. So what happened? Too much array map, filter, reduce!
- 27. Array functional programming issues • Iterating over arrays at each step • Allocating new arrays at each step • GC’ing those arrays
- 28. Array filter, map, reduce
- 29. RxJS Observables to the rescue! Stream Processing is nice
- 30. Observable Stream Processing • Data is only iterated over once • No intermediary arrays created • Less GC
- 31. Observable filter, map, reduce
- 32. Problem solved Less array allocation, less GC, less iterating over arrays
- 33. RxJS Observables Are about to solve another problem
- 34. Web Socket Connectivity Issues • If you walked between buildings you lost the socket connection • If you closed your lid for too long, you lost the socket connection • If the network hiccupped or went down, you lost the socket connection • If the server has an issue or restarts, you lose the socket connection • If you’re using 4G and your singal gets too low, you lose the socket connection
- 35. Basically you had to stay very, very still..
- 36. Multiplexing Web Sockets • Connect to the socket • Send a subscription message for each data stream you care about • Group the incoming messages by data stream • Send an unsubscribe message when you’re done with a stream • If all streams are disconnected, close the socket.
- 37. Reconnecting Multiplexed Web Sockets Is A Pain • Reconnect the socket • Resend all subscriptions for previously subscribed streams • If you’re offline, wait for an online event, and reconnect • Add a delay and retry again (exponential) • What if the user switches views and needs different data streams during this?
- 38. Observables Are Lazy • Execute code upon subscription to set up the underlying data stream • Execute code upon disposal to teardown the underlying data stream
- 39. So you can do things like • Setup a web socket on subscription • Close the web socket on disposal • Send an AJAX request • Abort the AJAX request on disposal • Setup an event listener • Automatically remove event listeners on disposal
- 40. This is nice… • Setup a web socket on subscription • Close the web socket on disposal
- 41. Observables can be retried or repeated Simply by subscribing more than once!
- 42. Even better! RxJS has operators: retry, retryWhen and repeat
- 43. Code Example…
- 44. Application Development Is about data flow
- 45. How does your data flow through your app? • What events cause variables to be produced • How are those variables used and for what? • Are you sure you’re containing side-effects?
- 46. RxJS makes this easier to reason about (with practice)
- 47. Thinking in streams • Imagine every variable (not consts) is a stream • To get it, you need to figure out where it comes from • It could come from other variables, or it could come from an event • Figure out your side effects, these are the exit points for your data flow
- 48. Thinking in streams var c = a + b; // do something with c
- 49. Thinking in streams var cStream = Observable.combineLatest(aStream, bStream, (a, b) => a + b); cStream.subscribe(c => { // do something with c });
- 50. Data Flow Entry and Exit Common entry points • User input • Network I/O • Properties set • Events handled Common exit points • State • Persistence • View Renders • Emitting events
- 51. RxJS entry and exit observable sources for entries do() and subscribe() for exits
- 52. The idea is to keep your data flow contained And control those side effects
- 53. Recap • Any number of values, any amount of time • Are lazy, don’t do anything until you subscribe • Embody the set up and tear down of their underlying data sources • Can retry/repeat • Use do() and subscribe() for side-effects • Observables were designed from the beginning to be cancellable
- 54. Resources RxJS 2 on GitHub: https://github.com/Reactive-Extensions/RxJS Talk Examples: https://github.com/blesh/rx-talk-may-2015 RxSocketSubject: https://github.com/blesh/RxSocketSubject RxJS 3 (coming soon): https://github.com/ReactiveX/RxJS Observable Spec: https://github.com/jhusain/observable-spec Async Generator Proposal: https://github.com/jhusain/asyncgenerator
- 55. Who to follow @headinthebox – Erik Meijer (Rx) @jhusain – Jafar Husain (TC39 Observable) @mattpodwysocki – Matt Podwysocki (RxJS) @jeffbcross – Jeff Cross (Angular) @victorsavkin – Victor Savkin (Angular) @benlesh – (me) @benjchristensen – Ben Christensen (RxJava)