Binding Protocols with Associated Objects - Protocol Oriented Programming with Swift - ios

i'm an iOS dev with a couple of years of experience with swift, but rarely i've used PAT's...
This time, I was trying to move some code from an app that i've developed to a shared library that I use in a couple of projects. The case is about a Factory that uses various Builders (that are decorators of my business resources) via an Abstract Builder protocol, to obtain Items (in the real case, ViewControllers).
The Builder relays upon some variables that the Factory passes to him, but those are at the application level, so, to extract this logic and put it into my library, i need to use a generic reference, and because I want to work in a Protocol Oriented Programming manner, it is an AssociatedType.
// The item that i want to receive from my factory
protocol Item {
var content: String { get }
}
// This is the Builder interface that the Factory consumes
protocol Builder {
// The Abstract Parameters that the Application should define
associatedtype Parameters
func build(_ parameters: Parameters) -> Item?
}
// The BusinessResource of my library
protocol BusinessResource { }
// The Factory that consumes the Builders
protocol Factory {
associatedtype FactoryBuilder: Builder
var parameters: FactoryBuilder.Parameters { get }
func make(from businessResource: BusinessResource) -> Item?
}
// The generic implementation of my Factory
extension Factory {
func make(from businessResource: BusinessResource) -> Item? {
guard let builder = businessResource as? FactoryBuilder else {
return nil
}
return builder.build(self.parameters)
}
}
At this point everything looks good.
I have two protocols and those are binded together, sharing a common type who is generic (the Builder Parameters).
So, on the application layer, now i could introduce my concrete Parameters (i'll call them ConcreteParameters XD)
// The concrete parameters of the Application Factory
struct ConcreteParameters {
let string: String
}
// The Builder interface restricting Parameters to ConcreteParameters
protocol BindedBuilder: Builder where Parameters == ConcreteParameters {
}
// The Factory interface restricting Parameters to ConcreteParameters
protocol BindedFactory: AbstractFactory where FactoryParameters: ConcreteParameters {
}
So far, so good. Everything looks in place and I'm start thinking that this could work, so now i try to implement a concrete Factory on the application to try if this really works.
// The concrete output of my Builder
struct ConcreteItem: Item {
var content: String
}
// The concrete BusinessResource that i get from my library
struct ConcreteObject: BusinessResource {
let string: String
}
// The decoration extension that makes ConcreteObject compliant with Builder
extension ConcreteObject: Builder {
typealias Parameters = ConcreteParameters
func build(_ parameters: ConcreteParameters) -> Item? {
return ConcreteItem(content: parameters.string + self.string)
}
}
// The real Factory inside my app
class ConcreteFactory: BindedFactory {
typealias FactoryBuilder = BindedBuilder
var parameters: ConcreteParameters {
return ConcreteParameters(string: "Hello ")
}
}
let item = ConcreteFactory().make(from: ConcreteObject(string: "world!"))
print(item ?? "NOT WORKING")
At this point something breaks... I get this error:
[EDIT: Error came from a previous version of the snippet, AbstractFactori is current Factory]
It is a Bug??
I really don't know how to solve this...

I think in this case you need to use a concrete type to alias FactoryBuilder instead of BindedBuilder, as protocols do not conform to themselves.
This code effectively compiles, would something like that match your requirements?
class ConcreteFactory: BindedFactory {
typealias FactoryBuilder = ConcreteObject
var parameters: ConcreteParameters {
return ConcreteParameters(string: "Hello ")
}
}
Otherwise you can also try type erasing BindedBuilder and create AnyBindedBuilder, as suggested in the same link.

Related

How to mock classes of external framework with delegates in iOS?

I am working in an iOS application called ConnectApp and I am using a framework called Connector. Now, Connector framework completes actual connection task with BLE devices and let my caller app (i.e. ConnectApp) know the connection request results through ConnectionDelegate. Let's see example code,
ConnectApp - host app
class ConnectionService: ConnectionDelegate {
func connect(){
var connector = Connector()
connector.setDelegate(self)
connector.connect()
}
func onConnected(result: ConnectionResult) {
//connection result
}
}
Connector Framework
public class ConnectionResult {
// many complicated custom variables
}
public protocol ConnectionDelegate {
func onConnected(result: ConnectionResult)
}
public class Connector {
var delegate: ConnectionDelegate?
func setDelegate(delegate: ConnectionDelegate) {
self.delegate = delegate
}
func connect() {
//…..
// result = prepared from framework
delegate?.onConnected(result)
}
}
Problem
Sometimes developers have no BLE device and we need to mock the Connector layer of framework. In case of simple classes (i.e. with simpler methods) we could have used inheritance and mock the Connector with a MockConnector which might override the lower tasks and return status from MockConnector class. But when I need to deal with a ConnectionDelegate which returns complicated object. How can I resolve this issue?
Note that framework does not provide interfaces of the classes rather we need to find way around for concrete objects like, Connector, ConnectionDelegate etc.
Update 1:
Trying to apply Skwiggs's answer so I created protocol like,
protocol ConnectorProtocol: Connector {
associatedType MockResult: ConnectionResult
}
And then injecting real/mock using strategy pattern like,
class ConnectionService: ConnectionDelegate {
var connector: ConnectorProtocol? // Getting compiler error
init(conn: ConnectorProtocol){
connector = conn
}
func connect(){
connector.setDelegate(self)
connector.connect()
}
func onConnected(result: ConnectionResult) {
//connection result
}
}
Now I am getting compiler error,
Protocol 'ConnectorProtocol' can only be used as a generic constraint because it has Self or associated type requirements
What am I doing wrong?
In Swift, the cleanest way to create a Seam (a separation that allows us to substitute different implementations) is to define a protocol. This requires changing the production code to talk to the protocol, instead of a hard-coded dependency like Connector().
First, create the protocol. Swift lets us attach new protocols to existing types.
protocol ConnectorProtocol {}
extension Connector: ConnectorProtocol {}
This defines a protocol, initially empty. And it says that Connector conforms to this protocol.
What belongs in the protocol? You can discover this by changing the type of var connector from the implicit Connector to an explicit ConnectorProtocol:
var connector: ConnectorProtocol = Connector()
Xcode will complain about unknown methods. Satisfy it by copying the signature of each method it needs into the protocol. Judging from your code sample, it may be:
protocol ConnectorProtocol {
func setDelegate(delegate: ConnectionDelegate)
func connect()
}
Because Connector already implements these methods, the protocol extension is satisfied.
Next, we need a way for the production code to use Connector, but for test code to substitute a different implementation of the protocol. Since ConnectionService creates a new instance when connect() is called, we can use a closure as a simple Factory Method. The production code can supply a default closure (creating a Connector) like with a closure property:
private let makeConnector: () -> ConnectorProtocol
Set its value by passing an argument to the initializer. The initializer can specify a default value, so that it makes a real Connector unless told otherwise:
init(makeConnector: (() -> ConnectorProtocol) = { Connector() }) {
self.makeConnector = makeConnector
super.init()
}
In connect(), call makeConnector() instead of Connector(). Since we don't have unit tests for this change, do a manual test to confirm we didn't break anything.
Now our Seam is in place, so we can begin writing tests. There are two types of tests to write:
Are we calling Connector correctly?
What happens when the delegate method is called?
Let's make a Mock Object to check the first part. It's important that we call setDelegate(delegate:) before calling connect(), so let's have the mock record all calls in an array. The array gives us a way to check the call order. Instead of having the test code examine the array of calls (acting as a Test Spy which just records stuff), your test will be cleaner if we make this a full-fledged Mock Object — meaning it will do its own verification.
final class MockConnector: ConnectorProtocol {
private enum Methods {
case setDelegate(ConnectionDelegate)
case connect
}
private var calls: [Methods] = []
func setDelegate(delegate: ConnectionDelegate) {
calls.append(.setDelegate(delegate))
}
func connect() {
calls.append(.connect)
}
func verifySetDelegateThenConnect(
expectedDelegate: ConnectionDelegate,
file: StaticString = #file,
line: UInt = #line
) {
if calls.count != 2 {
fail(file: file, line: line)
return
}
guard case let .setDelegate(delegate) = calls[0] else {
fail(file: file, line: line)
return
}
guard case .connect = calls[1] else {
fail(file: file, line: line)
return
}
if expectedDelegate !== delegate {
XCTFail(
"Expected setDelegate(delegate:) with \(expectedDelegate), but was \(delegate)",
file: file,
line: line
)
}
}
private func fail(file: StaticString, line: UInt) {
XCTFail("Expected setDelegate(delegate:) followed by connect(), but was \(calls)", file: file, line: line)
}
}
(That business with passing around file and line? This makes it so that any test failure will report the line that calls verifySetDelegateThenConnect(expectedDelegate:), instead of the line that calls XCTFail(_).)
Here's how you'd use this in ConnectionServiceTests:
func test_connect_shouldMakeConnectorSettingSelfAsDelegateThenConnecting() {
let mockConnector = MockConnector()
let service = ConnectionService(makeConnector: { mockConnector })
service.connect()
mockConnector.verifySetDelegateThenConnect(expectedDelegate: service)
}
That takes care of the first type of test. For the second type, there's no need to test that Connector calls the delegate. You know it does, and it's outside your control. Instead, write a test to call the delegate method directly. (You'll still want it to make a MockConnector to prevent any calls to the real Connector).
func test_onConnected_withCertainResult_shouldDoSomething() {
let service = ConnectionService(makeConnector: { MockConnector() })
let result = ConnectionResult(…) // Whatever you need
service.onConnected(result: result)
// Whatever you want to verify
}
You could try
protocol MockConnector: Connector {
associatedType MockResult: ConnectionResult
}
Then, for each connector you need to mock, define a concrete class that conforms to this mock connector
class SomeMockConnector: MockConnector {
struct MockResult: ConnectionResult {
// Any mocked variables for this connection result here
}
// implement any further requirements from the Connector class
var delegate: ConnectionDelegate?
func connect() {
// initialise your mock result with any specific data
let mockResult = MockResult()
delegate?.onConnected(mockResult)
}
}

Building composable objects in Swift with protocols

I'm trying to create a way to build compassable objects in Swift. I feel like I'm almost there with what I have but it's still not 100% correct.
What I'm aiming for is to have a FlowController object that can create our UIViewControllers and then give them any of the dependencies that they need.
What I'd also like to do is make this work as loosely as possible.
I have a small example here that works but is not ideal. I'll explain...
Here are two objects that can be used as components... Wallet and User.
class Wallet {
func topUp(amount: Int) {
print("Top up wallet with £\(amount)")
}
}
class User {
func sayHello() {
Print("Hello, world!")
}
}
We then define a Component enum that has cases for each of these...
enum Component {
case Wallet
case User
}
... And a protocol that defines a method requiresComponents that returns an array of Components.
This is where the problem arises. In order for the "factory object" to put the components into a Composable object we need to define the user and wallet properties in the protocol also.
protocol Composable {
var user: User? {get set}
var wallet: Wallet? {get set}
func requiresComponents() -> [Component]
}
In an attempt to make these properties "optional" (not Optional) I have defined an extension to the Composable protocol that defines these vars as nil.
extension Composable {
var user: User? {
get {return nil}
set {}
}
var wallet: Wallet? {
get {return nil}
set {}
}
}
Now I declare the class that I want to make Composable. As you can see it requires the User component and declares the variable.
class SomeComposableClass: Composable {
var user: User?
func requiresComponents() -> [Component] {
return [.User]
}
}
Now the FlowController that will create these and add the components to them. You can see here that I have had to take the object, create a local var version of it and then return the updated object. I think this is because it doesn't know the type of objects that will be conforming to the protocol so the parameter can't be mutated.
class FlowController {
func addComponents<T: Composable>(toComposableObject object: T) -> T {
var localObject = object
for component in object.requiresComponents() {
switch component {
case .Wallet:
localObject.wallet = Wallet()
print("Wallet")
case .User:
localObject.user = User()
print("User")
}
}
return localObject
}
}
Here I create the objects.
let flowController = FlowController()
let composable = SomeComposableClass()
And here I add the components. In production this would be done all inside the FlowController.
flowController.addComponents(toComposableObject: composable) // prints "User" when adding the user component
compassable.user?.sayHello() // prints "Hello, world!"
As you can see, it works here. The user object is added.
However, as you can also see. Because I have declared the vars in the protocol the composable object also has a reference to a wallet component (although it will always be nil).
composable.wallet // nil
I feel like I'm about 95% of the way there with this but what I'd like to be able to do is improve how the properties are declared. What I'd like is for that last line... composable.wallet to be a compile error.
I could do this by moving the declaration of the properties out of the protocol but then I have the problem of not being able to add the properties to any object that conforms to the Composable protocol.
What would be awesome is for the factory object to be able to add the properties without relying on the declaration. Or even have some sort of guard that says "if this object has a property call user then add the user component to it". Or something like that.
If anyone knows how I could get the other 5% of this working it would be awesome. Like I said, this works, just not in an ideal way.
Thanks :D
Hacky Edit
Hmm... As a quick tacky, horrible, "no-one-should-do-this" edit. I have changed my protocol extension to be like this...
extension Composable {
var user: User? {
get {fatalError("Access user")}
set {fatalError("Set user")}
}
var wallet: Wallet? {
get {fatalError("Access wallet")}
set {fatalError("Set waller")}
}
}
Now at least the program will crash if I try to access a variable I have not defined. But it's still not ideal.
Edit after reading Daniel's blog
OK, I think I've done what I wanted. Just not sure that it's exactly Swifty. Although, I also think it might be. Looking for a second opinion :)
So, my components and protocols have become this...
// these are unchanged
class Wallet {
func topUp(amount: Int) {
print("Top up wallet with £\(amount)")
}
}
// each component gets a protocol
protocol WalletComposing {
var wallet: Wallet? {get set}
}
class User {
func sayHello() {
print("Hello, world!")
}
}
protocol UserComposing {
var user: User? {get set}
}
Now the factory method has changed...
// this is the bit I'm unsure about.
// I now have to check for conformance to each protocol
// and add the components accordingly.
// does this look OK?
func addComponents(toComposableObject object: AnyObject) {
if var localObject = object as? UserComposing {
localObject.user = User()
print("User")
}
if var localObject = object as? WalletComposing {
localObject.wallet = Wallet()
print("Wallet")
}
}
This allows me to do this...
class SomeComposableClass: UserComposing {
var user: User?
}
class OtherClass: UserComposing, WalletComposing {
var user: User?
var wallet: Wallet?
}
let flowController = FlowController()
let composable = SomeComposableClass()
flowController.addComponents(toComposableObject: composable)
composable.user?.sayHello()
composable.wallet?.topUp(amount: 20) // this is now a compile time error which is what I wanted :D
let other = OtherClass()
flowController.addComponents(toComposableObject: other)
other.user?.sayHello()
other.wallet?.topUp(amount: 10)
This seems like a good case for applying the Interface Segregation Principle
Specifically, rather than having a master Composable protocol, have many smaller protocols like UserComposing and WalletComposing. Then your concrete types that wish to compose those various traits, would just list their "requiredComponents" as protocols they conform to, i.e:
class FlowController : UserComposing, WalletComposing
I actually wrote a blog post that talks about this more extensively and gives more detailed examples at http://www.danielhall.io/a-swift-y-approach-to-dependency-injection
UPDATE:
Looking at the updated question and sample code, I would only suggest the following refinement:
Going back to your original design, it might make sense to define a base Composing protocol that requires any conforming class to create storage for composed traits as a dictionary. Something like this:
protocol Composing : class {
var traitDictionary:[String:Any] { get, set }
}
Then, use protocol extensions to add the actual composable trait as a computed property, which reduces the boilerplate of having to create those properties in every conforming class. This way any class can conform to any number of trait protocols without having to declare a specific var for each. Here's a more complete example implementation:
class FlowController {
static func userFor(instance:UserComposing) -> User {
return User()
}
static func walletFor(instance:WalletComposing) -> Wallet {
return Wallet()
}
}
protocol Composing : class {
var traitDictionary:[String:Any] { get, set }
}
protocol UserComposing : Composing {}
extension UserComposing {
var user:User {
get {
if let user = traitDictionary["user"] as? User {
return user
}
else {
let user = FlowController.userFor(self)
traitDictionary["user"] = user
return user
}
}
}
}
protocol WalletComposing {}
extension WalletComposing {
var wallet:Wallet {
get {
if let wallet = traitDictionary["wallet"] as? Wallet {
return wallet
}
else {
let wallet = FlowController.walletFor(self)
traitDictionary["wallet"] = wallet
return wallet
}
}
}
}
class AbstractComposing {
var traitDictionary = [String:Any]()
}
Not only does this get rid of those pesky optionals you have to unwrap everywhere, but it makes the injection of user and wallet implicit and automatic. That means that your classes will already have the right values for those traits even inside their own initializers, no need to explicitly pass each new instance to an instance of FlowController every time.
For example, your last code snippet would now become simply:
class SomeComposableClass: AbstractComposing, UserComposing {} // no need to declare var anymore
class OtherClass: AbstractComposing, UserComposing, WalletComposing {} //no vars here either!
let composable = SomeComposableClass() // No need to instantiate FlowController and pass in this instance
composable.user.sayHello() // No unwrapping the optional, this is guaranteed
composable.wallet.topUp(amount: 20) // this is still a compile time error which is what you wanted :D
let other = OtherClass() // No need to instantiate FlowController and pass in this instance
other.user.sayHello()
other.wallet.topUp(amount: 10) // It all "just works" ;)

Where to put reusable functions in IOS Swift?

New to IOS programming but just wondering where is the best place to put functions that I would use throughout my code. For example, I want to write a few functions to perform a POST request to a web service and return a dictionary. Maybe another function to do some calculations. Is it best to create another .swift file and put all my functions there. And what would be a good name to give the file if so?
public func postRequest() -> [String:String] {
// do a post request and return post data
return ["someData" : "someData"]
}
The best way is to create a helper class with static functions, like this:
class Helper{
static func postRequest() -> [String:String] {
// do a post request and return post data
return ["someData" : "someData"]
}
}
Now every time you need to use postRequest you can just use like so: Helper.postRequest()
I usually create a separate class if I have functions that will be used by multiple classes, especially for the ones involving network operations.
If you just have separate functions that will be used, you can simply create static functions inside that class so it is easily accessible by other classes in a static way:
class DataController {
static func getData() -> [String:String] {
// do some operations
return ["someData" : "someData"]
}
}
let data = DataController.getData() // example
However, what often has been the case for me (especially if it involves more complicated operations) was that these network operations needed to establish an initial connection beforehand or required some initial setups, and they also performed asynchronous operations that needed to be controlled. If this is the case and you will often be calling such methods, you might want to create a singleton object that you could use throughout different classes and functions. This way, you could do the initial setup or establish an initial connection just once, and then do the rest as needed with the other functions, instead of doing them every time the function gets called.
Creating a singleton object is pretty simple in Swift:
class DataController {
static let sharedInstance = DataController() // singleton object
init() {
// do initial setup or establish an initial connection
}
func getData() -> [String:String] {
// do some operations
return ["someData" : "someData"]
}
}
let data = DataController.sharedInstance.getData() // example
For the name of the class, I usually name it something like DataController or DataHelper, but anything that makes sense as a "helper" class would work.
Hope this helps :)
For reusable functions it depends what I decide to use. For this specific case I use a separate file, because posting to a backend will become more complicated when the application evolves. In my app I use a backend class, with all kinds of helper classes:
struct BackendError {
var message : String
}
struct SuccessCall {
var json : JSON
var containsError : Bool {
if let error = json["error"].string {
return true
}
else {
return false
}
}
}
typealias FailureBlock = (BackendError) -> Void
typealias SuccessBlock = (SuccessCall) -> Void
typealias AlamoFireRequest = (path: String, method: Alamofire.Method, data: [String:String]) -> Request
typealias GetFunction = (path: String , data: [String : String], failureBlock: FailureBlock, successBlock: SuccessBlock) -> Void
class Backend {
func getRequestToBackend (token: String )(path: String , data: [String : String], failureBlock: FailureBlock, successBlock:
}
For other cases I often use extensions on Swift classes. Like for getting a random element from an Array.
extension Array {
func sampleItem() -> T {
let index = Int(arc4random_uniform(UInt32(self.count)))
return self[index]
}
}
This very old question but I would like to chirp some more points.
There are a few option, basically you can write your utility functions in Swift -
A class with static function. For example
class CommonUtility {
static func someTask() {
}
}
// uses
CommonUtility.someTask()
Also, you can have class method's as well instead of static method but those functions can be overridden by subclasses unlike static functions.
class CommonUtility {
class func someTask() {
}
}
// uses
CommonUtility.someTask()
Secondly, you can have Global functions as well, that are not part of any class and can be access anywhere from your app just by name.
func someTask() {
}
Though, selecting one over other is very subjective and I thing this is ok to make a class with static function in this particular case, where you need to achieve networking functionality but if you have some functions which perform only one task than Global function is a way to go because Global functions are more modular and separate out single tasks for a single function.
In case of static functions, if we access one of the static member, entire class gets loaded in memory. But in case of global function, only that particular function will be loaded in mem
You can create a separate swift class, might name it WebServicesManager.swift, and write all methods related to web requests in it.
You can use class methods, or singleton pattern to access the methods.

A static field inherited from the base class or protocol - how?

I want to be able to have the classes which have a static property (field) which is either inherited from the base class or "mixed" from a protocol. And every class should have it's own implementation of that property. Is it possible? Preferably, it to be immutable.
class C1 {
static let stProperty = "my prorepty1"
}
class C2 {
static let stProperty = "my prorepty2"
}
It's possible, but it's really hard to make this useful in Swift. How do you plan to refer to this property? Let's start with a super-simple implementation:
protocol SomeProtocol {
static var prop: String { get }
}
class C1: SomeProtocol {
static let prop = "This is One"
}
Great. So now I want a function that uses this:
func useProp(x: SomeProtocol) -> String {
return x.prop
// 'SomeProtocol' does not have a member named 'prop'
}
That doesn't work. x is an instance, but I want the type.
// Accessing members of protocol type value 'SomeProtocol.Type' is unimplemented
func useProp(x: SomeProtocol.Type) -> String {
return x.prop
}
This is probably how it will work some day given the word "unimplemented." But it doesn't work today.
func useProp(x: SomeProtocol) -> String {
// Accessing members of protocol type value 'SomeProtocol.Type' is unimplemented
return x.dynamicType.prop
}
Same thing.
Today, you really have to hang this on the object itself and not use static or class:
protocol SomeProtocol {
var prop: String { get }
}
class C1: SomeProtocol {
let prop = "This is One"
}
func useProp(x: SomeProtocol) -> String {
return x.prop
}
That's not so terrible in many cases, since the value for the class is probably also the value for any given instance of the class. And it's really all we can do today.
Of course your problem might be that you don't have an instance yet and you need this information to build an instance. That's really hard today and you should probably rethink your design. You'll generally have to use some other pattern like a Builder. See Generic Types Collection for more.
Now you also said:
or "mixed" from a protocol
I wouldn't say "mixed" here. If you really mean this like a Ruby "mixin", there is no such thing in Swift today. Swift folks often refer to this feature as "default implementation," and it's not currently possible (though I do expect it to come eventually). The only thing you can do in the protocol is say that the implementor has to provide this method somehow. You can't provide it for them.
Sure you can do that with a protocol:
protocol SomeProtocol {
static var foo: String { get }
}
class One: SomeProtocol {
class var foo: String {
get {
return "This is One"
}
}
}
Btw I agree with Rob Napier below that this is a bit off a oddball feature. I do think there are probably use-cases for it, but I also think those can be better implemented with other language features
protocol P {
class var stProperty: String { get }
}
class C1 {
class var stProperty: String {
return = "my property1"
}
}
class C2 {
class var stProperty: String {
return = "my property2"
}
}
Usage:
C2.prop //"my property2"
If you try:
C2.prop = "new value" //"cannot assign to the result of this expression"

How to use generic protocol as a variable type

Let's say I have a protocol :
public protocol Printable {
typealias T
func Print(val:T)
}
And here is the implementation
class Printer<T> : Printable {
func Print(val: T) {
println(val)
}
}
My expectation was that I must be able to use Printable variable to print values like this :
let p:Printable = Printer<Int>()
p.Print(67)
Compiler is complaining with this error :
"protocol 'Printable' can only be used as a generic constraint because
it has Self or associated type requirements"
Am I doing something wrong ? Anyway to fix this ?
**EDIT :** Adding similar code that works in C#
public interface IPrintable<T>
{
void Print(T val);
}
public class Printer<T> : IPrintable<T>
{
public void Print(T val)
{
Console.WriteLine(val);
}
}
//.... inside Main
.....
IPrintable<int> p = new Printer<int>();
p.Print(67)
EDIT 2: Real world example of what I want. Note that this will not compile, but presents what I want to achieve.
protocol Printable
{
func Print()
}
protocol CollectionType<T where T:Printable> : SequenceType
{
.....
/// here goes implementation
.....
}
public class Collection<T where T:Printable> : CollectionType<T>
{
......
}
let col:CollectionType<Int> = SomeFunctiionThatReturnsIntCollection()
for item in col {
item.Print()
}
As Thomas points out, you can declare your variable by not giving a type at all (or you could explicitly give it as type Printer<Int>. But here's an explanation of why you can't have a type of the Printable protocol.
You can't treat protocols with associated types like regular protocols and declare them as standalone variable types. To think about why, consider this scenario. Suppose you declared a protocol for storing some arbitrary type and then fetching it back:
// a general protocol that allows for storing and retrieving
// a specific type (as defined by a Stored typealias
protocol StoringType {
typealias Stored
init(_ value: Stored)
func getStored() -> Stored
}
// An implementation that stores Ints
struct IntStorer: StoringType {
typealias Stored = Int
private let _stored: Int
init(_ value: Int) { _stored = value }
func getStored() -> Int { return _stored }
}
// An implementation that stores Strings
struct StringStorer: StoringType {
typealias Stored = String
private let _stored: String
init(_ value: String) { _stored = value }
func getStored() -> String { return _stored }
}
let intStorer = IntStorer(5)
intStorer.getStored() // returns 5
let stringStorer = StringStorer("five")
stringStorer.getStored() // returns "five"
OK, so far so good.
Now, the main reason you would have a type of a variable be a protocol a type implements, rather than the actual type, is so that you can assign different kinds of object that all conform to that protocol to the same variable, and get polymorphic behavior at runtime depending on what the object actually is.
But you can't do this if the protocol has an associated type. How would the following code work in practice?
// as you've seen this won't compile because
// StoringType has an associated type.
// randomly assign either a string or int storer to someStorer:
var someStorer: StoringType =
arc4random()%2 == 0 ? intStorer : stringStorer
let x = someStorer.getStored()
In the above code, what would the type of x be? An Int? Or a String? In Swift, all types must be fixed at compile time. A function cannot dynamically shift from returning one type to another based on factors determined at runtime.
Instead, you can only use StoredType as a generic constraint. Suppose you wanted to print out any kind of stored type. You could write a function like this:
func printStoredValue<S: StoringType>(storer: S) {
let x = storer.getStored()
println(x)
}
printStoredValue(intStorer)
printStoredValue(stringStorer)
This is OK, because at compile time, it's as if the compiler writes out two versions of printStoredValue: one for Ints, and one for Strings. Within those two versions, x is known to be of a specific type.
There is one more solution that hasn't been mentioned on this question, which is using a technique called type erasure. To achieve an abstract interface for a generic protocol, create a class or struct that wraps an object or struct that conforms to the protocol. The wrapper class, usually named 'Any{protocol name}', itself conforms to the protocol and implements its functions by forwarding all calls to the internal object. Try the example below in a playground:
import Foundation
public protocol Printer {
typealias T
func print(val:T)
}
struct AnyPrinter<U>: Printer {
typealias T = U
private let _print: U -> ()
init<Base: Printer where Base.T == U>(base : Base) {
_print = base.print
}
func print(val: T) {
_print(val)
}
}
struct NSLogger<U>: Printer {
typealias T = U
func print(val: T) {
NSLog("\(val)")
}
}
let nsLogger = NSLogger<Int>()
let printer = AnyPrinter(base: nsLogger)
printer.print(5) // prints 5
The type of printer is known to be AnyPrinter<Int> and can be used to abstract any possible implementation of the Printer protocol. While AnyPrinter is not technically abstract, it's implementation is just a fall through to a real implementing type, and can be used to decouple implementing types from the types using them.
One thing to note is that AnyPrinter does not have to explicitly retain the base instance. In fact, we can't since we can't declare AnyPrinter to have a Printer<T> property. Instead, we get a function pointer _print to base's print function. Calling base.print without invoking it returns a function where base is curried as the self variable, and is thusly retained for future invocations.
Another thing to keep in mind is that this solution is essentially another layer of dynamic dispatch which means a slight hit on performance. Also, the type erasing instance requires extra memory on top of the underlying instance. For these reasons, type erasure is not a cost free abstraction.
Obviously there is some work to set up type erasure, but it can be very useful if generic protocol abstraction is needed. This pattern is found in the swift standard library with types like AnySequence. Further reading: http://robnapier.net/erasure
BONUS:
If you decide you want to inject the same implementation of Printer everywhere, you can provide a convenience initializer for AnyPrinter which injects that type.
extension AnyPrinter {
convenience init() {
let nsLogger = NSLogger<T>()
self.init(base: nsLogger)
}
}
let printer = AnyPrinter<Int>()
printer.print(10) //prints 10 with NSLog
This can be an easy and DRY way to express dependency injections for protocols that you use across your app.
Addressing your updated use case:
(btw Printable is already a standard Swift protocol so you’d probably want to pick a different name to avoid confusion)
To enforce specific restrictions on protocol implementors, you can constrain the protocol's typealias. So to create your protocol collection that requires the elements to be printable:
// because of how how collections are structured in the Swift std lib,
// you’d first need to create a PrintableGeneratorType, which would be
// a constrained version of GeneratorType
protocol PrintableGeneratorType: GeneratorType {
// require elements to be printable:
typealias Element: Printable
}
// then have the collection require a printable generator
protocol PrintableCollectionType: CollectionType {
typealias Generator: PrintableGenerator
}
Now if you wanted to implement a collection that could only contain printable elements:
struct MyPrintableCollection<T: Printable>: PrintableCollectionType {
typealias Generator = IndexingGenerator<T>
// etc...
}
However, this is probably of little actual utility, since you can’t constrain existing Swift collection structs like that, only ones you implement.
Instead, you should create generic functions that constrain their input to collections containing printable elements.
func printCollection
<C: CollectionType where C.Generator.Element: Printable>
(source: C) {
for x in source {
x.print()
}
}

Resources