Building multiplayer game using Reactive Streams

This talk introduces streams and reactive streams. You will learn how stream-based programming can be used to implement web frontend and multiplayer backend of the classic game: Snake.

Building dynamic applications using imperative approach tends to create lots of unmaintainable code. Stream-based programming tries to solve this problem by introducing fully declarative way of defining application logic. While using streams, you will focus on what needs to be done, not how and when.

The talk is divided into 3 parts. In the first part you will learn how to create a frontend of the Snake web game using streams as building blocks. Then, we will move to the server side and use Scala and Akka Stream library to create backend service that will allow the game to be played by multiple players. In the third part, we will discuss reactive streams and how they make asynchronous communication safe.

Transcript

1. KSUG / Kraków / 29.02.2016
   @miciek
   Building Multiplayer Game
   using Streams and Reactive Streams
   Michał
   Płachta
   

   View Slide

2. KSUG / Kraków / 29.02.2016
   @miciek
   Streams?
   ● way of defining application logic
   ● focused on data transformations
   ● unbounded collections
   ● have hardcoded time
   ● declarative
   

   View Slide

3. KSUG / Kraków / 29.02.2016
   @miciek
   What you will learn?
   ● how frontend web app written in
   streams look like
   ● how backend app written using
   streams look like
   ● how do Reactive Streams keep our
   streams safe?
   

   View Slide

4. KSUG / Kraków / 29.02.2016
   @miciek
   Meet Snake & Fruit
   Snake
   Fruit
   

   View Slide

5. KSUG / Kraków / 29.02.2016
   @miciek
   Our First Stream: “Ticks” stream
   

   View Slide

6. KSUG / Kraków / 29.02.2016
   @miciek
   “Directions” stream
   

   View Slide

7. KSUG / Kraków / 29.02.2016
   @miciek
   “Snake’s Head Positions” stream
   

   View Slide

8. KSUG / Kraków / 29.02.2016
   @miciek
   Stream operators: map & scan
   stream.map(f)
   ● transforms values from
   stream using function f
   ● returns a new stream
   with transformed values
   stream.scan(init, f)
   ● accumulates values
   from stream using
   function f
   ○ f(acc,val) =>
   new_acc
   ● returns a new stream
   that outputs
   accumulator on each
   new value
   

   View Slide

9. KSUG / Kraków / 29.02.2016
   @miciek
   Once again: the code
   

   View Slide

10. KSUG / Kraków / 29.02.2016
    @miciek
    Displaying values from the stream
    

    View Slide

11. KSUG / Kraków / 29.02.2016
    @miciek
    Streams are reusable
    

    View Slide

12. KSUG / Kraków / 29.02.2016
    @miciek
    Streams & Operators are lazy
    No elements
    =
    I don’t do
    anything
    

    View Slide

13. KSUG / Kraków / 29.02.2016
    @miciek
    “Key presses” stream
    

    View Slide

14. KSUG / Kraków / 29.02.2016
    @miciek
    “Left Key Presses” stream
    

    View Slide

15. KSUG / Kraków / 29.02.2016
    @miciek
    “Right Key Presses” stream
    

    View Slide

16. KSUG / Kraków / 29.02.2016
    @miciek
    Stream operators: filter
    stream.filter(p)
    ● checks each value from
    stream against the
    predicate passed as p
    ● returns a new stream
    that outputs only values
    that satisfy condition p
    Already used:
    ● map(f)
    ● scan(init, f(acc, val))
    

    View Slide

17. KSUG / Kraków / 29.02.2016
    @miciek
    “Left Rotations” stream
    

    View Slide

18. KSUG / Kraków / 29.02.2016
    @miciek
    “Right Rotations” stream
    

    View Slide

19. KSUG / Kraków / 29.02.2016
    @miciek
    “Actions” stream
    

    View Slide

20. KSUG / Kraków / 29.02.2016
    @miciek
    Stream operators: merge
    stream.merge(stream2)
    ● stream is merged with
    stream2
    ● returns a new stream that
    outputs values from both
    stream and stream2 as
    they appear
    Already used:
    ● map(f)
    ● scan(init, f(acc, val))
    ● filter(p)
    

    View Slide

21. KSUG / Kraków / 29.02.2016
    @miciek
    Recap: Current “Directions” stream
    

    View Slide

22. KSUG / Kraków / 29.02.2016
    @miciek
    “Directions” stream based on user input
    

    View Slide

23. KSUG / Kraków / 29.02.2016
    @miciek
    “Directions” stream: Better, but no timing!
    vs
    

    View Slide

24. KSUG / Kraków / 29.02.2016
    @miciek
    Stream operators: sampledBy
    stream.sampledBy(stream2)
    ● last value from stream is
    taken when any value
    appears in stream2
    ● returns a new stream that
    outputs values from
    stream sampled by values
    from stream2
    Already used:
    ● map(f)
    ● scan(init, f(acc, val))
    ● filter(p)
    ● merge(stream2)
    

    View Slide

25. KSUG / Kraków / 29.02.2016
    @miciek
    “Directions” stream: user input & timing
    

    View Slide

26. KSUG / Kraków / 29.02.2016
    @miciek
    “Snake Head Positions”: no change needed
    

    View Slide

27. KSUG / Kraków / 29.02.2016
    @miciek
    User moves the snake: the code
    

    View Slide

28. KSUG / Kraków / 29.02.2016
    @miciek
    Current implementation
    10 streams
    ● “ticks”
    ● “key presses”
    ● “left key presses”
    ● “right key presses”
    ● “left rotations”
    ● “right rotations”
    ● “actions”
    ● “direction changes”
    ● “directions”
    ● “snake head positions”
    + map, scan, filter, merge &
    sampledBy
    

    View Slide

29. KSUG / Kraków / 29.02.2016
    @miciek
    Stream operators: slidingWindow
    stream.slidingWindow(n)
    ● aggregate last n values
    ● returns a new stream that
    outputs arrays of last n
    values from the stream
    ● can be implemented using
    scan
    Already used:
    ● map(f)
    ● scan(init, f(acc, val))
    ● filter(p)
    ● merge(stream2)
    ● sampledBy(stream2)
    

    View Slide

30. KSUG / Kraków / 29.02.2016
    @miciek
    “Snakes” stream
    

    View Slide

31. KSUG / Kraków / 29.02.2016
    @miciek
    Quiz: Eating a fruit?
    6 operators
    ● map(f)
    ● scan(init, f(acc, val))
    ● filter(p)
    ● merge(stream2)
    ● sampledBy(stream2)
    ● slidingWindow(n)
    WHICH STREAMS AND OPERATORS
    SHOULD WE REUSE TO
    IMPLEMENT “EATING A FRUIT”
    FUNCTIONALITY?
    11 streams
    ● “ticks”
    ● “key presses”
    ● “left key presses”
    ● “right key presses”
    ● “left rotations”
    ● “right rotations”
    ● “actions”
    ● “direction changes”
    ● “directions”
    ● “snake head positions”
    ● “snakes”
    

    View Slide

32. KSUG / Kraków / 29.02.2016
    @miciek
    “Fruit eaten events” stream
    

    View Slide

33. KSUG / Kraków / 29.02.2016
    @miciek
    Frontend in action!
    

    View Slide

34. KSUG / Kraków / 29.02.2016
    @miciek
    Architecture: 11 streams, 2 ins & 2 outs
    

    View Slide

35. KSUG / Kraków / 29.02.2016
    @miciek
    Snake’s going multiplayer
    

    View Slide

36. KSUG / Kraków / 29.02.2016
    @miciek
    Snake Server App Requirements
    ● client must send snake positions to the server each time it
    changes
    ● server must generate fruits
    ● server must keep scores
    ● server must broadcast:
    ○ positions of all snakes to other players
    ○ positions of fruits to all players
    ○ current scoreboard to all players
    

    View Slide

37. KSUG / Kraków / 29.02.2016
    @miciek
    Let’s design Server App Flow!
    one player flow
    

    View Slide

38. KSUG / Kraków / 29.02.2016
    @miciek
    WebSocket connected to the Flow
    

    View Slide

39. KSUG / Kraków / 29.02.2016
    @miciek
    WebSocket as a stream: Server App
    

    View Slide

40. KSUG / Kraków / 29.02.2016
    @miciek
    Snake App Flow (one player)
    

    View Slide

41. KSUG / Kraków / 29.02.2016
    @miciek
    Let’s implement Server Flow for one player
    

    View Slide

42. KSUG / Kraków / 29.02.2016
    @miciek
    Moved to server: “Fruit eaten events” stream
    

    View Slide

43. KSUG / Kraków / 29.02.2016
    @miciek
    “Fruits” stream
    

    View Slide

44. KSUG / Kraków / 29.02.2016
    @miciek
    “Fruits” stream using Akka Stream
    

    View Slide

45. KSUG / Kraków / 29.02.2016
    @miciek
    “Score updates” stream
    

    View Slide

46. KSUG / Kraków / 29.02.2016
    @miciek
    “Scores” stream
    

    View Slide

47. KSUG / Kraków / 29.02.2016
    @miciek
    “Scores” stream using Akka Stream
    

    View Slide

48. KSUG / Kraków / 29.02.2016
    @miciek
    Game Logic Server Flow
    

    View Slide

49. KSUG / Kraków / 29.02.2016
    @miciek
    Game Logic Flow using Akka Stream
    

    View Slide

50. KSUG / Kraków / 29.02.2016
    @miciek
    Merging Player Flows outputs
    

    View Slide

51. KSUG / Kraków / 29.02.2016
    @miciek
    Server app flow using Akka Stream
    

    View Slide

52. KSUG / Kraków / 29.02.2016
    @miciek
    Backend in action!
    

    View Slide

53. KSUG / Kraków / 29.02.2016
    @miciek
    What if there are thousands of players?
    

    View Slide

54. KSUG / Kraków / 29.02.2016
    @miciek
    Crossing Async Boundary
    Push Model
    ● not safe when
    Subscriber slower
    ● perfect when
    Subscriber faster
    Pull Model
    ● too slow when
    Subscriber faster
    ● perfect when
    Subscriber slower
    

    View Slide

55. KSUG / Kraków / 29.02.2016
    @miciek
    Reactive Streams
    Streams
    with
    supply & demand
    ● start in Pull Mode
    ● Consumer asks for elements (= demand)
    ● Publisher goes into Push Mode
    ● supplies elements until demand is satisfied
    

    View Slide

56. KSUG / Kraków / 29.02.2016
    @miciek
    Reactive Streams
    Lack of demand
    =
    Backpressure
    

    View Slide

57. KSUG / Kraków / 29.02.2016
    @miciek
    Example of Reactive Streams
    

    View Slide

58. KSUG / Kraków / 29.02.2016
    @miciek
    Reactive Streams
    

    View Slide

59. KSUG / Kraków / 29.02.2016
    @miciek
    Reactive Streams API
    public interface Publisher {
    public void subscribe(Subscriber super T> s);
    }
    public interface Subscriber {
    public void onSubscribe(Subscription s);
    public void onNext(T t);
    public void onError(Throwable t);
    public void onComplete();
    }
    public interface Subscription {
    public void request(long n);
    public void cancel();
    }
    public interface Processor extends Subscriber, Publisher {}
    

    View Slide

60. KSUG / Kraków / 29.02.2016
    @miciek
    Simple API, hard implementation...
    https://github.com/reactive-streams/reactive-streams-jvm/blob/v1.0.0/README.md#specification
    

    View Slide

61. KSUG / Kraków / 29.02.2016
    @miciek
    Reactive Streams API implementations
    ● Project Reactor
    ● Ratpack
    ● Vert.x
    ● Slick
    ● Akka Stream
    ● JDK 9
    ● Integrations:
    ○ Kafka,
    ○ Mongo,
    ○ Cassandra
    

    View Slide

62. KSUG / Kraków / 29.02.2016
    @miciek
    Akka Stream model
    Publisher => Source
    Processor => Flow
    Subscriber => Sink
    Our Snake implementation uses
    Reactive Streams!
    

    View Slide

63. KSUG / Kraków / 29.02.2016
    @miciek
    ● NO
    ○ frontend is pushing every 100ms
    ● Possible solutions:
    ○ Reactive Streams web frameworks?
    ○ conflate/expand?
    ○ buffer between frontend &
    backend?
    Is Snake Multiplayer safe?
    

    View Slide

64. KSUG / Kraków / 29.02.2016
    @miciek
    Links
    github.com/miciek/web-snake-react-bacon Client Side Code
    github.com/miciek/snake-multiplayer-akka-streams Server Side Code
    rxmarbles.com/ Play with streams and operators!
    doc.akka.io/docs/akka-stream-and-http-experimental/2.0.3/ Akka Stream
    baconjs.github.io/ BaconJS - streams in JavaScript
    

    View Slide

65. KSUG / Kraków / 29.02.2016
    @miciek
    Building Multiplayer Game
    using Streams and Reactive Streams
    Michał Płachta
    www.michalplachta.com
    THANK YOU!
    

    View Slide