Functional Programming with Cats

Transcript

1. Functional Programming with Kats Lars Hupel February 17th, 2017
2. Functional Programming with Cats Lars Hupel February 17th, 2017
3. What is Cats? “ Lightweight, modular, and extensible library for functional programming ” 3
4. What’s in Cats? ▶ type classes ▶ data types ▶ monad transformers ▶ pretty syntax ▶ optimization macros ▶ laws 4
5. Ecosystem Alleycats Cats instances and classes which are outlaws, miscreants, and ne’er-do-wells Dogs Data structures for pure functional programming in Scala Kittens Automatic type class derivation for Cats 5
8. Agenda 1 Exercise 1: Error Handling 2 Exercise 2: Trees 3 Exercise 3: Free Monads 4 Exercise 4: Numerics 8
9. Agenda 1 Exercise 1: Error Handling 2 Exercise 2: Trees 3 Exercise 3: Free Monads 4 Exercise 4: Numerics 9
10. Exceptions Functional programmers don’t like exceptions. Instead: Either[Error, A] ... where Error is a custom type 10
11. Example def stringToInt(x: String): Either[String, Int] = { try { Right(x.toInt) } catch { case _: NumberFormatException => Left(s”Invalid number $x”) } } 11
12. Example def stringToInt(x: String): Either[String, Int] for { h <- stringToInt(hourField) m <- stringToInt(minField) } yield Time(h, m) 11
13. Accumulation def stringToInt(x: String): ValidatedNel[String, Int] = { try { Validated.valid(x.toInt) } catch { case _: NumberFormatException => Validated.invalidNel(s”Invalid number $x”) } } 12
14. Accumulation def stringToInt(x: String): ValidatedNel[String, Int] (stringToInt(hourField) |@| stringToInt(minField)) map { Time(h, m) } 12
16. Agenda 1 Exercise 1: Error Handling 2 Exercise 2: Trees 3 Exercise 3: Free Monads 4 Exercise 4: Numerics 14
17. http://abstrusegoose.com/467
18. Binary Trees Classic interview exercise 16
20. Agenda 1 Exercise 1: Error Handling 2 Exercise 2: Trees 3 Exercise 3: Free Monads 4 Exercise 4: Numerics 18
21. monad (n.) (in the pantheistic philosophy of Giordano Bruno) a fundamental metaphysical unit that is spatially extended and psychically aware – Collins English Dictionary
23. How to structure applications In functional programming, we separate pure from effectful code. ▶ in OO parlance: “Domain-driven design” 21
24. How to structure applications In functional programming, we separate pure from effectful code. ▶ in OO parlance: “Domain-driven design” ▶ But how do we structure effectful code? ▶ database access ▶ web requests ▶ terminal I/O ▶ ... 21
25. The “Free Monad” approach: I/O as a DSL What do we need to write a terminal application? ▶ read from standard input ▶ write to standard output ▶ open file ▶ read from file ▶ ... 22
26. The “Free Monad” approach: I/O as a DSL What do we need to write a terminal application? ▶ read from standard input ▶ write to standard output ▶ open file ▶ read from file ▶ ... 22
27. A datatype for terminals sealed trait Terminal[A] case object ReadLine extends Terminal[String] case class WriteLine(line: String) extends Terminal[Unit] 23
28. A datatype for terminals sealed trait Terminal[A] case object ReadLine extends Terminal[String] case class WriteLine(line: String) extends Terminal[Unit] ▶ Terminal is an ordinary class hierarchy ▶ nicely represents what we want ▶ But what to do with it? How do we compose it? 23
29. A datatype for terminals sealed trait Terminal[A] case object ReadLine extends Terminal[String] case class WriteLine(line: String) extends Terminal[Unit] ▶ Terminal is an ordinary class hierarchy ▶ nicely represents what we want ▶ But what to do with it? How do we compose it? 23 It looks like you need a monad. Want help with that?
30. Coyoneda? “ What is sometimes called the co-Yoneda lemma is a basic fact about presheaves (a basic fact of topos theory): it says that every presheaf is a colimit of representables and more precisely that it is the “colimit over itself of all the representables contained in it”. ” – nLab 24
31. Coyoneda? “ What is sometimes called the co-Yoneda lemma is a basic fact about presheaves (a basic fact of topos theory): it says that every presheaf is a colimit of representables and more precisely that it is the “colimit over itself of all the representables contained in it”. ” – nLab 24
34. Agenda 1 Exercise 1: Error Handling 2 Exercise 2: Trees 3 Exercise 3: Free Monads 4 Exercise 4: Numerics 27
35. Semigroup trait Semigroup[A] { def combine(x: A, y: A): A } 28
36. Semigroup trait Semigroup[A] { def combine(x: A, y: A): A } Law: Associativity combine(x, combine(y, z)) == com- bine(combine(x, y), z) 28
37. Monoids trait Monoid[A] { def combine(x: A, y: A): A def empty: A } Law: Neutral element combine(x, empty) == x 29
38. Monoidal structures Lots of things are monoids. ▶ Int, BigInt, ... 30
39. Monoidal structures Lots of things are monoids. ▶ Int, BigInt, ... ▶ List[T] 30
40. Monoidal structures Lots of things are monoids. ▶ Int, BigInt, ... ▶ List[T] ▶ Map[K, V] 30
41. Monoidal structures Lots of things are monoids. ▶ Int, BigInt, ... ▶ List[T] ▶ Map[K, V] ▶ ... 30
42. Algebraic hierarchy Semigroup Monoid Group 31
43. Algebraic hierarchy Semigroup Monoid Group AddSemigroup AddMonoid AddGroup 31
44. Algebraic hierarchy Semigroup Monoid Group AddSemigroup AddMonoid AddGroup MulSemigroup MulMonoid MulGroup 31
45. Algebraic hierarchy AddSemigroup AddMonoid AddGroup MulSemigroup MulMonoid MulGroup Semiring 31
46. Algebraic hierarchy AddSemigroup AddMonoid AddGroup MulSemigroup MulMonoid MulGroup Semiring AddAbGroup MulAbGroup Rig Rng Ring Field 31
48. Q & A  larsrh  larsr h