It seems that there are a few situations in which a required keyword on Swift class functions would be extremely beneficial, particularly due to the ability of class functions to return Self.
When returning Self from a class func, there are unfortunately two restrictions that make implementing said function very difficult/inhibitive:
You cannot use Self as a type check inside the function implementation, ie:
class Donut {
class func gimmeOne() -> Self {
// Compiler error, 'Self' is only available in a protocol or as the result of a class method
return Donut() as Self
}
}
You cannot return the actual type of the class itself, ie:
class Donut {
class func gimmeOne() -> Self {
// Compiler error, 'Donut' is not convertible to 'Self'
return Donut()
}
}
The reason for these compiler errors is valid. If you have a GlazedDonut subclass that does not override this class function, it is possible that calling GlazedDonut.gimmeOne() will give you back a Donut, which is not a Self.
It seems this situation could be alleviated by allowing classes to specify these functions with required. This would ensure that any subclasses override the method and encur their own round of type checking, making sure that a GlazedDonut returns itself in all cases, eliminating the possibility for a Donut to come back.
Is there a technical, authoritative reason why this has not been added? I'd like to suggest it as an improvement to the Swift team, but want to ensure there isn't an obvious reason why it has been omitted, or cannot be accomplished.
The idea for this question originates here:
https://stackoverflow.com/a/25924224/88111
required is generally only used on initializers, because initializers are not always inherited in Swift. Therefore, to allow you to call an initializer on a variable class (i.e. a value of metaclass type, say Foo.Type), you need to know that that class Foo, and all possible subclasses, have this initializer.
However, methods (both instance methods and class methods) are always inherited. Therefore, required is not necessary.
By the way, your assertion that "You cannot return the actual type of the class itself" is not true. (In fact, the error "'Self' is only available in a protocol or as the result of a class method" itself says you can return the type of the class itself.) Similar to in Objective-C, you can do:
class Donut {
required init() { }
class func gimmeOne() -> Self {
return self()
}
}
You could use a protocol to make the method 'required'
protocol IDonut{
class func gimmeOne()->Donut;
}
class Donut:IDonut {
class func gimmeOne() -> Donut {
return Donut();
}
}
class GlazedDonut: Donut, IDonut{
override class func gimmeOne()->Donut{
return GlazedDonut();
}
}
Related
Essentially both are used to modify behaviour of a function to our custom needs. But why the necessity arise to have two ways to do the same thing when both are used for same purpose.
I'm assuming, if a function has HEAD which takes parameters and a BODY which has a certain functionality with those parameters
Mutating function is used when we have to modify at the HEAD.
Mutating -> HEAD -> parameters and
Overriding function is used when we have to modify at the BODY
Overriding -> BODY -> functionality
I've searched over the internet..but found no satisfactory explanation anywhere. Please help me understand them better. Please correct me if I'm wrong.
Mutating
Swift structs are immutable objects meaning that you cannot change its properties within its functions. You need to explicitly mention that you agree to make changes to its properties by adding the mutating keyword in the function definition. However this mutating jargon is required only for value types in Swift - structs and enums.
struct MutatingExample {
var number: Int = 0
// Add 'mutating' to resolve the error
func changeNumber(changedNumber: Int) {
self.number = changedNumber // Error: Cannot assign to property: 'self' is immutable
}
}
Here is an useful post that might provide you more insights - What does the Swift 'mutating' keyword mean?
Reference types such as class do just fine and allow you to change the properties within their functions.
Override
Override is a concept used in inheritance. By that we can infer that override is applicable to reference types such as class and value type(struct/enums) are out of question.
As the name implies, we use the keyword to override an existing functionality, typically that of a super class. For example,
class Parent {
func getName() {
print("Parent")
}
}
class Child: Parent {
// Add override to resolve error
func getName() {
print("Child") // Error: Overriding declaration requires an 'override' keyword
}
}
Useful link: https://www.hackingwithswift.com/sixty/8/3/overriding-methods
I have a parent class with a method signature like this:
class MyParent {
func doStuff(property: Self) {
fatalError("Not Implemented")
}
}
I then inherit form it like so:
final class MyInherited: MyParent {
var arbitraryVal:String = "Hello"
override func doStuff(property: MyInherited) {
self.arbitraryVal = property.arbitraryVal
}
}
However, it says that doStuff doesn't override any method from its superclass. Is there a way to do what I'm trying to do here and have a parent class required a method property be of type Self?
This violates LSP. Anything the superclass can do, the subclass must also be able to do. If MyParent.doStuff can accept a MyParent, then MyInherited.doStuff must also be able to accept a MyParent. Otherwise, if you passed this object to a function that accepted a MyParent, how would it type-check a later call to doStuff?
A subclass can restrict what it returns, but it cannot restrict what it accepts. So it's ok for the output of doStuff to be Self, but it's not possible for the input to be Self. The input has to be at least as permissive as its parent.
I would expect that you're receiving an error in MyParent, not just MyInherited. It should say:
Covariant 'Self' can only appear as the type of a property, subscript or method result; did you mean 'MyParent'?
And this is correct. It's not possible to restrict a parameter to a smaller set of values. This isn't a Swift limitation. This is a limitation of any sound type system that includes inheritance.
You can use a protocol instead of a superclass
protocol MyParent {
var arbitraryVal: String { get set }
func doStuff(property: MyParent)
}
If I declare
public class A: NSObject {
public class X { }
public init?(x: X? = nil) { }
}
all is fine. When using it like let a = A(), the initializer is called as expected.
Now, I'd like to have the nested class X private, and the parameterized init as well (has to be, of course). But a simple init?() should stay publicly available as it was before. So I write
public class B: NSObject {
private class X { }
private init?(x: X?) { }
public convenience override init?() { self.init(x: nil) }
}
But this gives an error with the init?() initializer: failable initializer 'init()' cannot override a non-failable initializer with the overridden initializer being the public init() in NSObject.
How comes I can effectively declare an initializer A.init?() without the conflict but not B.init?()?
Bonus question: Why am I not allowed to override a non-failable initializer with a failable one? The opposite is legal: I can override a failable initializer with a non-failable, which requires using a forced super.init()! and thus introduces the risk of a runtime error. To me, letting the subclass have the failable initializer feels more sensible since an extension of functionality introduces more chance of failure. But maybe I am missing something here – explanation greatly appreciated.
This is how I solved the problem for me:
I can declare
public convenience init?(_: Void) { self.init(x: nil) }
and use it like
let b = B(())
or even
let b = B()
— which is logical since its signature is (kind of) different, so no overriding here. Only using a Void parameter and omitting it in the call feels a bit strange… But the end justifies the means, I suppose. :-)
After a bit of fiddling I think I understand. Let's consider a protocol requiring this initializer and a class implementing it:
protocol I {
init()
}
class A : I {
init() {}
}
This gives the error: "Initializer requirement 'init()' can only be satisfied by a required initializer in non-final class 'A'". This makes sense, as you could always declare a subclass of A that doesn't inherit that initializer:
class B : A {
// init() is not inherited
init(n: Int) {}
}
So we need to make our initializer in A required:
class A : I {
required init() {}
}
Now if we look at the NSObject interface we can see that the initializer is not required:
public class NSObject : NSObjectProtocol {
[...]
public init()
[...]
}
We can confirm this by subclassing it, adding a different initializer and trying to use the normal one:
class MyObject : NSObject {
init(n: Int) {}
}
MyObject() // Error: Missing argument for parameter 'n:' in call
Now here comes the weird thing: We can extend NSObject to conform to the I protocol, even though it doesn't require this initializer:
extension NSObject : I {} // No error (!)
I honestly think this is either a bug or a requirement for ObjC interop to work (EDIT: It's a bug and already fixed in the latest version). This error shouldn't be possible:
extension I {
static func get() -> Self { return Self() }
}
MyObject.get()
// Runtime error: use of unimplemented initializer 'init()' for class '__lldb_expr_248.MyObject'
Now to answer your actual question:
In your second code sample, the compiler is right in that you cannot override a non-failable with a failable initializer.
In the first one, you aren't actually overriding the initializer (no override keyword either), but instead declaring a new one by which the other one can't be inherited.
Now that I wrote this much I'm not even sure what the first part of my answer has to do with your question, but it's nice to find a bug anyways.
I suggest you to do this instead:
public convenience override init() { self.init(x: nil)! }
Also have a look at the Initialization section of the Swift reference.
I want to use a strategy pattern to register a set of objects that implement a protocol. When I set this up, I get a compile error when trying to set the delegate that is part of the protocol.
For discussion purposes, I have slightly reworked the DiceGame from the Swift eBook's Delegation chapter. The changes of significance are:
protocol DiceGame - declares a delegate
class SnakesAndLadders implements DiceGame (and therefore the protocol and delegate)
class Games holds 3 instances of SnakesAndLadders as
1) a concrete class of SnakesAndLadders
2) a 'let' constant of protocol DiceGame
3) a 'var' variable of protocol DiceGame
We can set the delegate fine if we use the concrete class (snakesAndLadders). However, there is a compile error if we use 'let' to hold it as a protocol (diceGameAsLet) but it compiles if we hold the variable as a 'var' (diceGameAsVar).
It is easy to work around, however, the delegate itself never changes so should be held as a 'let' constant, as it is only the internal property that changes. I must not understand something (possibly subtle but significant) about protocols and how they work and should be used.
class Dice
{
func roll() -> Int
{
return 7 // always win :)
}
}
protocol DiceGame
{
// all DiceGames must work with a DiceGameDelegate
var delegate:DiceGameDelegate? {get set}
var dice: Dice {get}
func play()
}
protocol DiceGameDelegate
{
func gameDidStart( game:DiceGame )
func gameDidEnd( game:DiceGame )
}
class SnakesAndLadders:DiceGame
{
var delegate:DiceGameDelegate?
let dice = Dice()
func play()
{
delegate?.gameDidStart(self)
playGame()
delegate?.gameDidEnd(self)
}
private func playGame()
{
print("Playing the game here...")
}
}
class Games : DiceGameDelegate
{
let snakesAndLadders = SnakesAndLadders()
// hold the protocol, not the class
let diceGameAsLet:DiceGame = SnakesAndLadders()
var diceGameAsVar:DiceGame = SnakesAndLadders()
func setupDelegateAsClass()
{
// can assign the delegate if using the class
snakesAndLadders.delegate = self
}
func setupDelegateAsVar()
{
// if we use 'var' we can assign the delegate
diceGameAsVar.delegate = self
}
func setupDelegateAsLet()
{
// DOES NOT COMPILE - Why?
//
// We are not changing the dice game so want to use 'let', but it won't compile
// we are changing the delegate, which is declared as 'var' inside the protocol
diceGameAsLet.delegate = self
}
// MARK: - DiceGameDelegate
func gameDidStart( game:DiceGame )
{
print("Game Started")
}
func gameDidEnd( game:DiceGame )
{
print("Game Ended")
}
}
DiceGame is a heterogeneous protocol that you're using as a type; Swift will treat this type as a value type, and hence (just as for a structures), changing its mutable properties will mutate also the instance of the protocol type itself.
If you, however, add the : class keyword to the DiceGame protocol, Swift will treat it as a reference type, allowing you to mutate members of instances of it, without mutating the instance itself. Note that this will constraint the protocol as conformable to only by class types.
protocol DiceGame: class { ... }
With the addition of the above, the mutation of immutable diceGameAsLet:s properties will be allowed.
In this context, it's worth mentioning that the : class keyword is usually used to constrain protocols used as delegates (e.g., DiceGameDelegate in your example) as conformable to only by class types. With this additional constraint, the delegates can be used as types to which the delegate owner (e.g. some class) only hold a weak reference, useful in contexts where a strong reference to the delegate could create a retain cycle.
See e.g. the 2nd part of this answer for details.
The issue is that when you store something as a Protocol, even if it is a class, swift considers them to be a value type, instead of the reference type you are expecting them to be. Therefore, no part of it is allowed to be changed. Take a look at this reference for more information.
I just started to learn Swift (v. 2.x) because I'm curious how the new features play out, especially the protocols with Self-requirements.
The following example is going to compile just fine, but causes arbitrary runtime effects to happen:
// The protocol with Self requirement
protocol Narcissistic {
func getFriend() -> Self
}
// Base class that adopts the protocol
class Mario : Narcissistic {
func getFriend() -> Self {
print("Mario.getFriend()")
return self;
}
}
// Intermediate class that eliminates the
// Self requirement by specifying an explicit type
// (Why does the compiler allow this?)
class SuperMario : Mario {
override func getFriend() -> SuperMario {
print("SuperMario.getFriend()")
return SuperMario();
}
}
// Most specific class that defines a field whose
// (polymorphic) access will cause the world to explode
class FireFlowerMario : SuperMario {
let fireballCount = 42
func throwFireballs() {
print("Throwing " + String(fireballCount) + " fireballs!")
}
}
// Global generic function restricted to the protocol
func queryFriend<T : Narcissistic>(narcissistic: T) -> T {
return narcissistic.getFriend()
}
// Sample client code
// Instantiate the most specific class
let m = FireFlowerMario()
// The call to the generic function is verified to return
// the same type that went in -- 'FireFlowerMario' in this case.
// But in reality, the method returns a 'SuperMario' and the
// call to 'throwFireballs' will cause arbitrary
// things to happen at runtime.
queryFriend(m).throwFireballs()
You can see the example in action on the IBM Swift Sandbox here.
In my browser, the output is as follows:
SuperMario.getFriend()
Throwing 32 fireballs!
(instead of 42! Or rather, 'instead of a runtime exception', as this method is not even defined on the object it is called on.)
Is this a proof that Swift is currently not type-safe?
EDIT #1:
Unpredictable behavior like this has to be unacceptable.
The true question is, what exact meaning the keyword Self (capital first letter) has.
I couldn't find anything online, but there are at least these two possibilities:
Self is simply a syntactic shortcut for the full class name it appears in, and it could be substituted with the latter without any change in meaning. But then, it cannot have the same meaning as when it appears inside a protocol definition.
Self is a sort of generic/associated type (in both protocols and classes) that gets re-instantiated in deriving/adopting classes. If that is the case, the compiler should have refused the override of getFriend in SuperMario.
Maybe the true definition is neither of those. Would be great if someone with more experience with the language could shed some light on the topic.
Yes, there seems to be a contradiction. The Self keyword, when used as a return type, apparently means 'self as an instance of Self'. For example, given this protocol
protocol ReturnsReceived {
/// Returns other.
func doReturn(other: Self) -> Self
}
we can't implement it as follows
class Return: ReturnsReceived {
func doReturn(other: Return) -> Self {
return other // Error
}
}
because we get a compiler error ("Cannot convert return expression of type 'Return' to return type 'Self'"), which disappears if we violate doReturn()'s contract and return self instead of other. And we can't write
class Return: ReturnsReceived {
func doReturn(other: Return) -> Return { // Error
return other
}
}
because this is only allowed in a final class, even if Swift supports covariant return types. (The following actually compiles.)
final class Return: ReturnsReceived {
func doReturn(other: Return) -> Return {
return other
}
}
On the other hand, as you pointed out, a subclass of Return can 'override' the Self requirement and merrily honor the contract of ReturnsReceived, as if Self were a simple placeholder for the conforming class' name.
class SubReturn: Return {
override func doReturn(other: Return) -> SubReturn {
// Of course this crashes if other is not a
// SubReturn instance, but let's ignore this
// problem for now.
return other as! SubReturn
}
}
I could be wrong, but I think that:
if Self as a return type really means 'self as an instance of
Self', the compiler should not accept this kind of Self requirement
overriding, because it makes it possible to return instances which
are not self; otherwise,
if Self as a return type must be simply a placeholder with no further implications, then in our example the compiler should already allow overriding the Self requirement in the Return class.
That said, and here any choice about the precise semantics of Self is not bound to change things, your code illustrates one of those cases where the compiler can easily be fooled, and the best it can do is generate code to defer checks to run-time. In this case, the checks that should be delegated to the runtime have to do with casting, and in my opinion one interesting aspect revealed by your examples is that at a particular spot Swift seems not to delegate anything, hence the inevitable crash is more dramatic than it ought to be.
Swift is able to check casts at run-time. Let's consider the following code.
let sm = SuperMario()
let ffm = sm as! FireFlowerMario
ffm.throwFireballs()
Here we create a SuperMario and downcast it to FireFlowerMario. These two classes are not unrelated, and we are assuring the compiler (as!) that we know what we are doing, so the compiler leaves it as it is and compiles the second and third lines without a hitch. However, the program fails at run-time, complaining that it
Could not cast value of type
'SomeModule.SuperMario' (0x...) to
'SomeModule.FireFlowerMario' (0x...).
when trying the cast in the second line. This is not wrong or surprising behaviour. Java, for example, would do exactly the same: compile the code, and fail at run-time with a ClassCastException. The important thing is that the application reliably crashes at run-time.
Your code is a more elaborate way to fool the compiler, but it boils down to the same problem: there is a SuperMario instead of a FireFlowerMario. The difference is that in your case we don't get a gentle "could not cast" message but, in a real Xcode project, an abrupt and terrific error when calling throwFireballs().
In the same situation, Java fails (at run-time) with the same error we saw above (a ClassCastException), which means it attempts a cast (to FireFlowerMario) before calling throwFireballs() on the object returned by queryFriend(). The presence of an explicit checkcast instruction in the bytecode easily confirms this.
Swift on the contrary, as far as I can see at the moment, does not try any cast before the call (no casting routine is called in the compiled code), so a horrible, uncaught error is the only possible outcome. If, instead, your code produced a run-time "could not cast" error message, or something as gracious as that, I would be completely satisfied with the behaviour of the language.
The issue here seems to be a violation in contract:
You define getFriend() to return an instance of receiver (Self). The problem here is that SuperMario does not return self but it returns a new instance of type SuperMario.
Now, when FireFlowerMario inherits that method the contract says that the method should return a FireFlowerMario but instead, the inherited method returns a SuperMario! This instance is then treated as if it were a FireFlowerMario, specifically: Swift tries to access the instance variable fireballCount which does not exist on SuperMario and you get garbage instead.
You can fix it like this:
class SuperMario : Mario {
required override init() {
super.init()
}
override func getFriend() -> SuperMario {
print("SuperMario.getFriend()")
// Dynamically create new instance of the same type as the receiver.
let myClass = self.dynamicType
return myClass.init()
}
}
Why does the compiler allow it? It has a hard time catching something like this, I guess. For SuperMario, the contract is still valid: the method getFriend does return an instance of the same class. The contract breaks when you create the subclass FireFlowerMario: should the compiler notice that a superclass might violate the contract? This would be an expensive check and probably more suited for a static analyzer, IMHO (Also, what happens if the compiler doesn't have access to SuperMario's source? What happens if that class is from a library?)
So it's actually SuperMario's duty to ensure that the contract is still valid when subclassing.