Domain Driven Design with the F# type System -- F#unctional Londoners 2014

Domain Driven Design with the F# type System -- F#unctional Londoners 2014

Statically typed functional programming languages like F# encourage a very different way of thinking about types. The type system is your friend, not an annoyance, and can be used in many ways that might not be familiar to OO programmers.

Types can be used to represent the domain in a fine-grained, self documenting way. And in many cases, types can even be used to encode business rules so that you literally cannot create incorrect code. You can then use the static type checking almost as an instant unit test — making sure that your code is correct at compile time.

In this talk, we'll look at some of the ways you can use types as part of a domain driven design process, with some simple real world examples in F#. No jargon, no maths, and no prior F# experience necessary.

Code, links to video, etc., at http://fsharpforfunandprofit.com/ddd

NEW AND IMPROVED - added sections on:
* why OO, not FP is scary
* designing with states and transitions

26743454d396b65db61e38ccc833ecdb?s=128

Scott Wlaschin

March 13, 2014
Tweet

Transcript

  1. 1.

    type Contact = { FirstName: string MiddleInitial: string LastName: string

    EmailAddress: string IsEmailVerified: bool } // true if ownership of // email address is confirmed Domain Driven Design with the F# type system Prologue: how many things are wrong?
  2. 2.

    type Contact = { FirstName: string MiddleInitial: string LastName: string

    EmailAddress: string IsEmailVerified: bool } Prologue: which values are optional?
  3. 3.

    type Contact = { FirstName: string MiddleInitial: string LastName: string

    EmailAddress: string IsEmailVerified: bool } Prologue: what are the constraints?
  4. 4.

    type Contact = { FirstName: string MiddleInitial: string LastName: string

    EmailAddress: string IsEmailVerified: bool } Prologue: what groups are atomic?
  5. 5.

    type Contact = { FirstName: string MiddleInitial: string LastName: string

    EmailAddress: string IsEmailVerified: bool } Prologue: domain logic?
  6. 6.

    Prologue: F# can help type Contact = { FirstName: string

    MiddleInitial: string LastName: string EmailAddress: string IsEmailVerified: bool }
  7. 7.

    Domain Driven Design with the F# type system Scott Wlaschin

    @ScottWlaschin fsharpforfunandprofit.com FPbridge.co.uk /ddd
  8. 10.

    What I’m going talk about: • Demystifying functional programming •

    Functional programming for real world applications • F# vs. C# for domain driven design • Understanding the F# type system • Designing with types
  9. 17.

    Functional programming is... ... good for mathematical and scientific tasks

    ... good for complicated algorithms ... really good for parallel processing ... but you need a PhD in computer science 
  10. 19.

    Must haves for BLOBA development... • Express requirements clearly •

    Rapid development cycle • High quality deliverables • Fun
  11. 20.
  12. 22.

    How do you implement a Value object? Equality based on

    comparing all properties PersonalName: FirstName = "Alice" LastName = "Adams" PersonalName: FirstName = "Alice" LastName = "Adams" Equal  Therefore must be immutable
  13. 23.

    class PersonalName { public PersonalName(string firstName, string lastName) { this.FirstName

    = firstName; this.LastName = lastName; } public string FirstName { get; private set; } public string LastName { get; private set; } } Value object definition in C#
  14. 24.

    class PersonalName { public PersonalName(string firstName, string lastName) { this.FirstName

    = firstName; this.LastName = lastName; } public string FirstName { get; private set; } public string LastName { get; private set; } } Value object definition in C#
  15. 25.

    class PersonalName { // all the code from above, plus...

    public override int GetHashCode() { return this.FirstName.GetHashCode() + this.LastName.GetHashCode(); } public override bool Equals(object other) { return Equals(other as PersonalName); } public bool Equals(PersonalName other) { if ((object) other == null) { return false; } return FirstName == other.FirstName && LastName == other.LastName; } Value object definition in C# (extra code for equality)
  16. 28.

    How do you implement an Entity object? Equality based on

    some sort of id Person: Id = 1 Name = "Alice Adams" Person: Id = 1 Name = "Bilbo Baggins" Equal X  Generally has mutable content
  17. 29.

    class Person { public Person(int id, PersonalName name) { this.Id

    = id; this.Name = name; } public int Id { get; private set; } public PersonalName Name { get; set; } } Entity object definition in C# (part 1)
  18. 30.

    class Person { // all the code from above, plus...

    public override int GetHashCode() { return this.Id.GetHashCode(); } public override bool Equals(object other) { return Equals(other as Person); } public bool Equals(Person other) { if ((object) other == null) { return false; } return Id == other.Id; } } Entity object definition in C# (part 2)
  19. 31.

    [<CustomEquality; NoComparison>] type Person = {Id:int; Name:PersonalName} with override this.GetHashCode()

    = hash this.Id override this.Equals(other) = match other with | :? Person as p -> (this.Id = p.Id) | _ -> false Entity object definition in F# with equality override
  20. 32.

    Entity object definition in F# with no equality allowed [<CustomEquality;

    NoComparison>] type Person = {Id:int; Name:PersonalName} [<NoEquality; NoComparison>]
  21. 33.

    type Person = { ... ... ... } let tryCreatePerson

    name = // validate on construction // if input is valid return something // if input is not valid return error   Advantages of immutability
  22. 34.

    class PersonalName { public PersonalName(string firstName, string lastName) { this.FirstName

    = firstName; this.LastName = lastName; } public string FirstName { get; private set; } public string LastName { get; private set; } public override int GetHashCode() { return this.FirstName.GetHashCode() + this.LastName.GetHashCode(); } public override bool Equals(object other) { return Equals(other as PersonalName); } public bool Equals(PersonalName other) { if ((object) other == null) { return false; } return FirstName == other.FirstName && LastName == other.LastName; } } Reviewing the C# code so far... class Person { public Person(int id, PersonalName name) { this.Id = id; this.Name = name; } public int Id { get; private set; } public PersonalName Name { get; set; } public override int GetHashCode() { return this.Id.GetHashCode(); } public override bool Equals(object other) { return Equals(other as Person); } public bool Equals(Person other) { if ((object) other == null) { return false; } return Id == other.Id; } } : IValue : IEntity
  23. 35.

    [<StructuralEquality;NoComparison>] type PersonalName = { FirstName : string; LastName :

    string } Reviewing the F# code so far... [<NoEquality; NoComparison>] type Person = { Id : int; Name : PersonalName }
  24. 36.

    Comparing C# vs. F# C# F# Value objects? Non-trivial Easy

    Entity objects? Non-trivial Easy Value objects by default? No Yes Immutable objects by default? No Yes Can you tell Value objects from Entities at a glance? No Yes Understandable by non-programmer? No Yes C# vs. F# for DDD
  25. 41.

    Communication in DDD: “Bounded Context” Business Chemistry un-ionize unionize Supermarket

    Email System Spam Spam Sales Warehouse Product Product Marketing Finance Customer Customer
  26. 43.

    Communication in DDD: “Ubiquitous Language” Chemistry Ion Atom Molecule Polymer

    Compound Bond Sales Product Promotion Customer Tracking
  27. 44.

    Communication in DDD: “Ubiquitous Language” Chemistry Ion Atom Molecule Polymer

    Compound Bond Sales Product Promotion Customer Tracking Warehouse Product Stock Transfer Depot Tracking
  28. 45.

    module CardGame = type Suit = Club | Diamond |

    Spade | Heart type Rank = Two | Three | Four | Five | Six | Seven | Eight | Nine | Ten | Jack | Queen | King | Ace type Card = Suit * Rank type Hand = Card list type Deck = Card list type Player = {Name:string; Hand:Hand} type Game = {Deck:Deck; Players: Player list} type Deal = Deck –› (Deck * Card) type PickupCard = (Hand * Card) –› Hand '*' means a pair. Choose one from each type Ubiquitous language list type is built in
  29. 46.

    module CardGame = type Suit = Club | Diamond |

    Spade | Heart type Rank = Two | Three | Four | Five | Six | Seven | Eight | Nine | Ten | Jack | Queen | King | Ace type Card = Suit * Rank type Hand = Card list type Deck = Card list type Player = {Name:string; Hand:Hand} type Game = {Deck:Deck; Players: Player list} type Deal = Deck –› (Deck * Card) type PickupCard = (Hand * Card) –› Hand
  30. 47.

    module CardGame = type Suit = Club | Diamond |

    Spade | Heart type Rank = Two | Three | Four | Five | Six | Seven | Eight | Nine | Ten | Jack | Queen | King | Ace type Card = Suit * Rank type Hand = Card list type Deck = Card list type Player = {Name:string; Hand:Hand} type Game = {Deck:Deck; Players: Player list} type Deal = Deck –› (Deck * Card) type PickupCard = (Hand * Card) –› Hand
  31. 48.
  32. 51.

    Creating new types New types are constructed by combining other

    types using two basic operations: type typeW = typeX "times" typeY type typeZ = typeX "plus" typeY
  33. 57.

    Using tuples for data × = Alice, Jan 12th Bob,

    Feb 2nd Carol, Mar 3rd Set of people Set of dates
  34. 58.

    Using tuples for data × = Alice, Jan 12th Bob,

    Feb 2nd Carol, Mar 3rd Set of people Set of dates type Birthday = Person * Date
  35. 60.

    Representing a choice + = 98˚ F 99˚ F 100˚

    F 101˚ F 37.0˚ C 37.5˚ C 38.0˚ C 38.5˚ C Temp F Temp C or
  36. 61.

    Representing a choice + = 98˚ F 99˚ F 100˚

    F 101˚ F 37.0˚ C 37.5˚ C 38.0˚ C 38.5˚ C Temp F Temp C or Tag these with “C” type Temp = | F of int | C of float
  37. 62.

    Using choices for data type PaymentMethod = | Cash |

    Cheque of int | Card of CardType * CardNumber
  38. 63.

    Working with a choice type type PaymentMethod = | Cash

    | Cheque of int | Card of CardType * CardNumber let printPayment method = match method with | Cash –› printfn “Paid in cash" | Cheque checkNo –› printfn “Paid by cheque: %i" checkNo | Card (cardType,cardNo) –› printfn “Paid with %A %A" cardType cardNo
  39. 64.

    Using choices vs. inheritance interface IPaymentMethod {..} class Cash :

    IPaymentMethod {..} class Cheque : IPaymentMethod {..} class Card : IPaymentMethod {..} type PaymentMethod = | Cash | Cheque of int | Card of CardType * CardNumber class Evil : IPaymentMethod {..} Data and code is scattered around many locations What goes in here? What is the common behaviour? OO version:
  40. 65.

    Summary: What are types for in FP? An annotation to

    a value for type checking type AddOne: int –› int Domain modelling tool type Deal = Deck –› (Deck * Card)
  41. 68.

    Required vs. Optional type PersonalName = { FirstName: string; MiddleInitial:

    string; LastName: string; } required required optional
  42. 69.

    Null is not the same as “optional” Length string –›

    int “a” “b” “c” 1 2 3 “a” “b” “c” null
  43. 71.

    Null is not the same as “optional” Length string –›

    int “a” “b” “c” null 1 2 3
  44. 72.
  45. 74.

    A better way for optional values + = “a” “b”

    “c” “a” “b” “c” missing or Tag with “Nothing” type OptionalString = | SomeString of string | Nothing
  46. 75.

    type OptionalInt = | SomeInt of int | Nothing type

    OptionalString = | SomeString of string | Nothing type OptionalBool = | SomeBool of bool | Nothing Defining optional types
  47. 76.

    The built-in “Option” type type PersonalName = { FirstName: string

    MiddleInitial: string LastName: string } type Option<'T> = | Some of 'T | None
  48. 77.

    The built-in “Option” type type PersonalName = { FirstName: string

    MiddleInitial: Option<string> LastName: string } type Option<'T> = | Some of 'T | None
  49. 78.

    The built-in “Option” type type PersonalName = { FirstName: string

    MiddleInitial: string option LastName: string } type Option<'T> = | Some of 'T | None
  50. 79.

    Single choice types type Something = | ChoiceA of A

    type Email = | Email of string type CustomerId = | CustomerId of int
  51. 80.

    Wrapping primitive types Is an EmailAddress just a string? Is

    a CustomerId just a int? Use single choice types to keep them distinct type EmailAddress = EmailAddress of string type PhoneNumber = PhoneNumber of string type CustomerId = CustomerId of int type OrderId = OrderId of int
  52. 81.

    Creating the EmailAddress type let createEmailAddress (s:string) = if Regex.IsMatch(s,@"^\S+@\S+\.\S+$")

    then (EmailAddress s) else ? let createEmailAddress (s:string) = if Regex.IsMatch(s,@"^\S+@\S+\.\S+$") then Some (EmailAddress s) else None createEmailAddress: string –› EmailAddress createEmailAddress: string –› EmailAddress option
  53. 82.

    Constrained strings type String50 = String50 of string let createString50

    (s:string) = if s.Length <= 50 then Some (String50 s) else None createString50 : string –› String50 option
  54. 84.

    Constrained numbers type OrderLineQty = OrderLineQty of int let createOrderLineQty

    qty = if qty >0 && qty <= 99 then Some (OrderLineQty qty) else None createOrderLineQty: int –› OrderLineQty option
  55. 85.

    type Contact = { FirstName: string MiddleInitial: string LastName: string

    EmailAddress: string IsEmailVerified: bool } The challenge, revisited
  56. 86.

    The challenge, revisited type Contact = { FirstName: string MiddleInitial:

    string option LastName: string EmailAddress: string IsEmailVerified: bool }
  57. 87.

    The challenge, revisited type Contact = { FirstName: String50 MiddleInitial:

    String1 option LastName: String50 EmailAddress: EmailAddress IsEmailVerified: bool }
  58. 88.
  59. 89.

    type Contact = { Name: PersonalName Email: EmailContactInfo } The

    challenge, revisited type PersonalName = { FirstName: String50 MiddleInitial: String1 option LastName: String50 } type EmailContactInfo = { EmailAddress: EmailAddress IsEmailVerified: bool }
  60. 90.

    Encoding domain logic Rule 1: If the email is changed,

    the verified flag must be reset to false. Rule 2: The verified flag can only be set by a special verification service type EmailContactInfo = { EmailAddress: EmailAddress IsEmailVerified: bool }
  61. 91.

    Encoding domain logic type VerifiedEmail = VerifiedEmail of EmailAddress type

    EmailContactInfo = | Unverified of EmailAddress | Verified of VerifiedEmail type VerificationService = (EmailAddress * VerificationHash) –› VerifiedEmail option
  62. 92.

    type EmailAddress = ... type VerifiedEmail = VerifiedEmail of EmailAddress

    type EmailContactInfo = | Unverified of EmailAddress | Verified of VerifiedEmail The challenge, completed type PersonalName = { FirstName: String50 MiddleInitial: String1 option LastName: String50 } type Contact = { Name: PersonalName Email: EmailContactInfo }
  63. 93.

    Making illegal states unrepresentable type Contact = { Name: Name

    Email: EmailContactInfo Address: PostalContactInfo }
  64. 94.

    Making illegal states unrepresentable type Contact = { Name: Name

    Email: EmailContactInfo Address: PostalContactInfo } New rule: “A contact must have an email or a postal address”
  65. 95.

    Making illegal states unrepresentable type Contact = { Name: Name

    Email: EmailContactInfo option Address: PostalContactInfo option } New rule: “A contact must have an email or a postal address”
  66. 96.

    Making illegal states unrepresentable “A contact must have an email

    or a postal address” implies: • email address only, or • postal address only, or • both email address and postal address
  67. 97.

    Making illegal states unrepresentable type ContactInfo = | EmailOnly of

    EmailContactInfo | AddrOnly of PostalContactInfo | EmailAndAddr of EmailContactInfo * PostalContactInfo type Contact = { Name: Name ContactInfo : ContactInfo } “A contact must have an email or a postal address”
  68. 98.

    Making illegal states unrepresentable type Contact = { Name: Name

    Email: EmailContactInfo Address: PostalContactInfo } type Contact = { Name: Name ContactInfo : ContactInfo } type ContactInfo = | EmailOnly of EmailContactInfo | AddrOnly of PostalContactInfo | EmailAndAddr of EmailContactInfo * PostalContactInfo AFTER: Email and address merged into one type “A contact must have an email or a postal address” BEFORE: Email and address separate
  69. 99.
  70. 100.

    Making illegal states unrepresentable type Contact = { Name: Name

    PrimaryContactInfo: ContactInfo SecondaryContactInfo: ContactInfo option } “A contact must have an email or a postal address” “A contact must have at least one way of being contacted” type ContactInfo = | Email of EmailContactInfo | Addr of PostalContactInfo
  71. 102.

    States and transitions State A State B State C Transition

    from A to B States and transitions Transition from B to A Transition from B to C
  72. 103.

    States and transitions Unverified EmailAddress Verified EmailAddress Verified States and

    transitions for email address Rule: "You can't send a verification message to a verified email" Rule: "You can't send a password reset message to a unverified email "
  73. 104.

    States and transitions Undelivered Out for delivery Delivered Put on

    truck Address not found Signed for States and transitions for shipments Rule: "You can't put a package on a truck if it is already out for delivery" Rule: "You can't sign for a package that is already delivered"
  74. 105.

    States and transitions Empty Cart Active Cart Paid Cart Add

    Item Remove Item Pay Add Item Remove Item States and transitions for shopping cart Rule: "You can't remove an item from an empty cart" Rule: "You can't change a paid cart" Rule: "You can't pay for a cart twice"
  75. 106.

    States and transitions Empty Cart Active Cart Paid Cart Add

    Item Remove Item Pay Add Item Remove Item States and transitions for shopping cart type ActiveCartData = { UnpaidItems: Item list } type PaidCartData = { PaidItems: Item list; Payment: Payment } no data needed
  76. 107.

    Modelling the shopping cart example type ActiveCartData = { UnpaidItems:

    Item list } type PaidCartData = { PaidItems: Item list; Payment: Payment} type ShoppingCart = | EmptyCart // no data | ActiveCart of ActiveCartData | PaidCart of PaidCartData Empty Cart Active Cart Paid Cart Add Item Remove Item Pay Add Item Remove Item
  77. 108.

    Shopping cart example initCart : Item –› ShoppingCart addToActive: (ActiveCartData

    * Item) –› ShoppingCart removeFromActive: (ActiveCartData * Item) –› ShoppingCart pay: (ActiveCartData * Payment) –› ShoppingCart Shopping Cart API Empty Cart Active Cart Paid Cart Add Item Remove Item Pay Add Item Remove Item
  78. 109.

    Shopping cart example let initCart item = { UnpaidItems=[item] }

    let addToActive (cart:ActiveCart) item = { cart with UnpaidItems = item :: cart.existingItems } Server code to add an item
  79. 110.

    Shopping cart example Client code to add an item using

    the API let addItem cart item = match cart with | EmptyCart –› initCart item | ActiveCart activeData –› addToActive(activeData,item) | PaidCart paidData –› ???
  80. 111.

    Shopping cart example let removeFromActive (cart:ActiveCart) item = let remainingItems

    = removeFromList cart.existingItems item match remainingItems with | [ ] –› EmptyCart | _ –› {cart with UnpaidItems = remainingItems} Server code to remove an item
  81. 112.

    Shopping cart example Client code to remove an item using

    the API let removeItem cart item = match cart with | EmptyCart –› ??? | ActiveCart activeData –› removeFromActive(activeData,item) | PaidCart paidData –› ???
  82. 113.

    Why design with state transitions? • Each state can have

    different allowable data. • All states are explicitly documented. • All transitions are explicitly documented. • It is a design tool that forces you to think about every possibility that could occur. Undelivered Out for delivery Delivered Put on truck Address not found Signed for
  83. 114.

    Review What I covered in this talk: • Ubiquitous language

    – Self-documenting designs • Algebraic types – products and sums • Designing with types – Options instead of null – Single case unions – Choices rather than inheritance – Making illegal states unrepresentable • States and transitions
  84. 115.

    Stuff I haven’t had time to cover: • Services •

    CQRS • The functional approach to use cases • Domain events • Error handling • And much more...
  85. 116.

    F# is low risk F# is the safe choice for

    functional-first development credit: @7sharp9 @MattDrivenDev Enterprise development F# on over 2 billion devices Mobile development Need to persuade your manager? -> FPbridge.co.uk/why-fsharp.html
  86. 117.

    Domain Driven Design with the F# type system DDD in

    F# resources fsharpforfunandprofit.com/ddd gorodinski.com tomasp.net/blog/type-first-development.aspx/ #fsharp on Twitter Contact me @ScottWlaschin FPbridge.co.uk