Breaking up with RxJava (Kotlin Coroutines & Channels)

Held on 26th June 2018 Kotlin Zagreb Meetup:
https://www.meetup.com/Kotlin-Zagreb/events/251748671/

When RxJava came along, it made our lives so easy that most Android developers jumped ship.
Soon, it was everywhere and everything was a stream. But was that really a good idea?
Now that we have Kotlin, a lot of painpoints solved by (ab)using RxJava are solvable by using plain Kotlin (well, experimental) coroutines.
So should we be jumping ship again? What makes Kotlin a better solution for most of your problems?
In this talk we’ll take a dive into some of Kotlin’s coroutine features and find out how, when and why to say “No” to RxJava.

Transcript

1. View Slide

2. View Slide

3. View Slide

4. Breaking up with Rx
   

   View Slide

5. Breaking up with Rx
   Ian Rumac
   Software Architect @ Undabot
   
   

   View Slide

6. The story so far
   

   View Slide

7. What RxJava gives us:

   The Good Parts
   

   View Slide

8. What RxJava gives us:

   The Good Parts
   • Easy stream manipulation
   • Simple concurrency
   • Chained operators
   • Observing state changes
   • Combining operators
   

   View Slide

9. What RxJava gives us:

   The Good Parts
   • Easy stream manipulation
   • Simple concurrency
   • Chained operators
   • Observing state changes
   • Combining operators
   

   View Slide

10. And we subscribed({});
    

    View Slide

11. And all was good.
    

    View Slide

12. View Slide

13. View Slide

14. View Slide

15. View Slide

16. View Slide

17. View Slide

18. View Slide

19. What RxJava gives us:

    The Bad Parts
    

    View Slide

20. • One library that spans across all layers

    (remember migrating?)
    • One library which you shape your code around
    • One paradigm you force everything into
    • One naming style
    • One extra layer of complexity
    • Code
    • Mental
    What RxJava gives us:

    The Bad Parts
    

    View Slide

21. • One library that spans across all layers

    (remember migrating?)
    • One library which you shape your code around
    • One paradigm you force everything into
    • One naming style
    • One extra layer of complexity
    • Code
    • Mental
    What RxJava gives us:

    The Bad Parts
    

    View Slide

22. The core
    

    View Slide

23. • Is your business logic wrapped in Rx?
    • Is your business logic a stream? 

    (not a stream of one )
    • Why do you represent the core logic of your app
    with a stream if it isn’t one?
    The core
    

    View Slide

24. So, what can we do about it?
    

    View Slide

25. Replacing the operator magic
    

    View Slide

26. Replacing the operator magic
    • Collection extensions already in stdlib
    

    View Slide

27. Replacing the operator magic
    • Collection extensions already in stdlib
    

    View Slide

28. Replacing the operator magic
    • Collection extensions already in stdlib
    

    View Slide

29. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    

    View Slide

30. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    

    View Slide

31. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    

    View Slide

32. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    • Receiver lambdas
    

    View Slide

33. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    • Receiver lambdas
    

    View Slide

34. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    • Receiver lambdas
    

    View Slide

35. Replacing the operator magic
    • Collection extensions already in stdlib
    • Easy to write your own
    • Receiver lambdas
    

    View Slide

36. Observing state changes
    

    View Slide

37. Observing state changes
    • Observable delegate
    

    View Slide

38. Observing state changes
    • Observable delegate
    

    View Slide

39. Observing state changes
    • Observable delegate
    • Vetoable delegate
    

    View Slide

40. Observing state changes
    • Observable delegate
    • Vetoable delegate
    

    View Slide

41. Observing state changes
    • Observable delegate
    • Vetoable delegate
    

    View Slide

42. Observing state changes
    • Observable delegate
    • Vetoable delegate
    

    View Slide

43. Observing state changes
    • Observable delegate
    • Vetoable delegate
    • Simple
    

    View Slide

44. Observing state changes
    • Observable delegate
    • Vetoable delegate
    • Simple
    • Cover a lot of usecases
    

    View Slide

45. Observing state changes
    • Observable delegate
    • Vetoable delegate
    • Simple
    • Cover a lot of usecases
    

    View Slide

46. Observing state changes
    • Observable delegate
    • Vetoable delegate
    • Simple
    • Cover a lot of usecases
    

    View Slide

47. Coroutines
    

    View Slide

48. Coroutines
    • Language feature
    • Lightweight
    • Context(s)
    • Easy thread switching
    

    View Slide

49. Coroutines
    

    View Slide

50. So just tldr light threads?
    

    View Slide

51. Coroutines
    

    View Slide

52. Coroutines
    • Easy concurrency
    • Yield!
    • Nesting and composing
    • cancelChildren()
    

    View Slide

53. Composing & Combining
    • Easy
    • Human readable
    • Wrap it any way you want
    

    View Slide

54. View Slide

55. doOne
    doTwo
    

    View Slide

56. View Slide

57. doOne doTwo
    

    View Slide

58. View Slide

59. doOne doTwo
    

    View Slide

60. View Slide

61. View Slide

62. View Slide

63. View Slide

64. But wait, there’s more!
    

    View Slide

65. Channels 101
    

    View Slide

66. Channels 101
    • A hot asynchronous stream of elements
    

    View Slide

67. Channels 101
    • A hot asynchronous stream of elements
    • Rendezvous transfer
    

    View Slide

68. Channels 101
    • A hot asynchronous stream of elements
    • Rendezvous transfer
    • Suspending
    

    View Slide

69. Channels 101
    • A hot asynchronous stream of elements
    • Rendezvous transfer
    • Suspending
    • Once it’s consumed, it’s consumed
    

    View Slide

70. Channels 101
    • A hot asynchronous stream of elements
    • Rendezvous transfer
    • Suspending
    • Once it’s consumed, it’s consumed
    • 1 item at a time
    

    View Slide

71. Channels 101
    

    View Slide

72. Channels 101
    

    View Slide

73. Channels 101
    

    View Slide

74. Channels 101
    

    View Slide

75. Channels 101
    

    View Slide

76. Channels 101
    

    View Slide

77. Channels of all shapes and sizes!
    • ArrayChannel - Buffered!
    • ConflatedChannel - Only the latest element
    • Ticker channel - Time based operations
    • BroadcastChannel & ArrayBroadcastChannel
    “Broadcast channel is a non-blocking primitive for communication between the sender and multiple receivers
    that subscribe for the elements using [open] function and unsubscribe using [SubscriptionReceiveChannel.close]
    function.”
    ReceiveChannel
    

    View Slide

78. View Slide

79. … well I still need my operators…
    

    View Slide

80. … well I still need my operators…
    

    View Slide

81. … well I still need my operators…
    

    View Slide

82. … well I still need my operators…
    

    View Slide

83. … well I still need my operators…
    

    View Slide

84. … well I still need my operators…
    

    View Slide

85. … well I still need my operators…
    

    View Slide

86. … well I still need my operators…
    

    View Slide

87. View Slide

88. …that’s all nice, but how do I test it?
    

    View Slide

89. …that’s all nice, but how do I test it?
    Just like any other code.
    (Context = Unconfined)
    

    View Slide

90. Thinking about dumping your old library?
    

    View Slide

91. Thinking about dumping your old library?
    • KotlinX Coroutines Rx 1&2
    • Kotlin Coroutine (Experimental) Retrofit Adapter
    • Still experimental!
    • Wrap the boilerplate
    

    View Slide

92. Conclusion
    

    View Slide

93. Conclusion
    Kotlin > RxJava
    

    View Slide

94. Conclusion
    

    View Slide

95. Conclusion
    Q & A?
    

    View Slide

96. View Slide

97. Thank you for your time!
    Ian Rumac
    Software Architect @Undabot
    

    View Slide