This is one of those things that seems simple enough, but doesn't work as you'd expect.
I'm working on a 'fluent/chaining'-style API for my classes to allow you to set properties via functions which can be chained together so you don't have to go crazy with initializers. Plus, it makes it more convenient when working with functions like map, filter and reduce which share the same kind of API.
Consider this RowManager extension...
extension RowManager
{
#discardableResult
public func isVisible(_ isVisible:Bool) -> RowManager
{
self.isVisible = isVisible
return self
}
}
This works exactly as one would expect. But there's a problem here... if you're working with a subclass of RowManager, this downcasts the object back to RowManager, losing all of the subclass-specific details.
"No worries!" I thought. "I'll just use Self and self to handle the type!" so I changed it to this...
extension RowManager
{
#discardableResult
public func isVisible(_ isVisible:Bool) -> Self // Note the capitalization representing the type, not instance
{
self.isVisible = isVisible
return self // Note the lowercase representing the instance, not type
}
}
...but that for some reason won't even compile giving the following error...
Command failed due to signal: Segmentation fault: 11
UPDATE
Doing more research, this seems to be because our code both is in, and also uses, dynamic libraries. Other questions here on SO also talk about that specific error in those cases. Perhaps this is a bug with the compiler because as others have correctly pointed out, this code works fine in a stand-alone test but as soon as the change is made in our code, the segmentation fault shows up.
Remembering something similar with class functions that return an instance of that type, I recalled how you had to use a private generic function to do the actual cast, so I tried to match that pattern with the following...
extension RowManager
{
#discardableResult
public func isVisible(_ isVisible:Bool) -> Self // Note the capitalization
{
self.isVisible = isVisible
return getTypedSelf()
}
}
private func getTypedSelf<T:RowManager>() -> T
{
guard let typedSelfInstance = self as? T
else
{
fatalError() // Technically this should never be reachable.
}
return typedSelfInstance
}
Unfortunately, that didn't work either.
For reference, here's the class-based code I attempted to base that off of (className is another extension that simply returns the string-representation of the name of the class you called it on)...
extension UITableViewCell
{
/// Attempts to dequeue a UITableViewCell from a table, implicitly using the class name as the reuse identifier
/// Returns a strongly-typed optional
class func dequeue(from tableView:UITableView) -> Self?
{
return self.dequeue(from:tableView, withReuseIdentifier:className)
}
/// Attempts to dequeue a UITableViewCell from a table based on the specified reuse identifier
/// Returns a strongly-typed optional
class func dequeue(from tableView:UITableView, withReuseIdentifier reuseIdentifier:String) -> Self?
{
return self.dequeue_Worker(tableView:tableView, reuseIdentifier:reuseIdentifier)
}
// Private implementation
private class func dequeue_Worker<T:UITableViewCell>(tableView:UITableView, reuseIdentifier:String) -> T?
{
return tableView.dequeueReusableCell(withIdentifier: reuseIdentifier) as? T
}
}
At WWDC Apple confirmed this was a Swift compiler issue that something else In our codebase was triggering, adding there should never be a case where you get a Seg11 fault in the compiler under any circumstances, so this question is actually invalid. Closing it now, but I will report back if they ever address it.
Related
This one has me stumped. I can't figure out why Swift is complaining that self.init is called more than once in this code:
public init(body: String) {
let parser = Gravl.Parser()
if let node = parser.parse(body) {
super.init(document: self, gravlNode: node)
} else {
// Swift complains with the mentioned error on this line (it doesn't matter what "whatever" is):
super.init(document: self, gravlNode: whatever)
}
}
Unless I'm missing something, it's very obvious that it is only calling init once. Funnily enough if I comment out the second line Swift complains that Super.init isn't called on all paths, lol.
What am I missing?
Update:
Ok so the problem was definitely trying to pass self in the call to super.init. I totally forgot I was doing that, ha. I think I had written that experimentally and gotten it to compile and thought that it might actually work, but looks like it's actually a bug that it compiled that way at all.
Anyhow, since passing self to an initializer is kind of redundant since it's the same object, I changed the parent initializer to accept an optional document parameter (it's just an internal initializer so no big deal) and if it's nil I just set it to self in the parent initializer.
For those curious, this is what the parent initializer (now) looks like:
internal init(document: Document?, gravlNode: Gravl.Node) {
self.value = gravlNode.value
super.init()
self.document = document ?? self as! Document
// other stuff...
}
I suspect this is a bad diagnostic (i.e the wrong error message). It would be very helpful if you had a full example we could experiment with, but this line doesn't make sense (and I suspect is the underlying problem):
super.init(document: self, gravlNode: node)
You can't pass self to super.init. You're not initialized yet (you're not initialized until you've called super.init). For example, consider the following simplified code:
class S {
init(document: AnyObject) {}
}
class C: S {
public init() {
super.init(document: self)
}
}
This leads to error: 'self' used before super.init call which I believe is the correct error.
EDIT: I believe Hamish has definitely uncovered a compiler bug. You can exploit it this way in Xcode 8.3.1 (haven't tested on 8.3.2 yet):
class S {
var x: Int
init(document: S) {
self.x = document.x
}
}
class C: S {
public init() {
super.init(document: self)
}
}
let c = C() // malloc: *** error for object 0x600000031244: Invalid pointer dequeued from free list
In my Swift code, I have a few methods that look like this:
protocol EditorDelegate {
// ...
func didStartSearch(query: String) -> Bool
}
class Editor: UIViewController {
func search(sender: AnyObject) {
let wasHandled = self.delegate?.didStartSearch(query) ?? false
if !wasHandled {
// do default searching behavior
}
}
}
This works, but it's not self-documenting. The didStartSearch method doesn't really communicate that it's returning a flag indicating whether the default behavior should be skipped.
What are some good options for handling this?
I tried to think of a better way to name the delegate function, below is what I came up with that makes the most sense to me. Enums are very powerful in swift, using them can make code more understandable at a glance. Also I like to think of things in terms of different states, which can be clearly defined with enums.
enum SearchStatus {
case Ready
case Processing
case Complete(result: String)
}
protocol EditorDelegate: class {
func statusForSearch(with query: String) -> SearchStatus
}
class Editor: UIViewController {
weak var delegate: EditorDelegate?
func search(sender: AnyObject) {
let query = "some input.."
guard let status = delegate?.statusForSearch(with: query) else {
// delegate not set
return
}
switch status {
case .Ready:
// actually do your search
break
case .Processing:
// search in progress
break
case .Complete(let result):
// do something with result
print("result = \(result)")
break
}
}
}
Think about the relationship between the delegating object and the object being delegated to, and the "message" (in high-level semantics, not just code) that a call to the delegate method is about. That is, does a delegate method call mean, "hey, in case you're interested, this thing happened," or, "this thing happened, what should I do about it?". If the delegate returns a value to the delegating object, you're probably in the second category.
Per Coding Guidelines for Cocoa, delegate methods should:
Start by identifying the delegating object (e.g. the tableView in tableView:didSelectRow: or the window in windowWillClose:). This is important because delegation can be a many-to-one relationship — if you have a delegate property (or setDelegate method) on one class, there's nothing preventing one object from being the delegate of many others, and indeed that can be a useful pattern for whoever adopts your delegate protocol.
If the method asks something of the delegate, either name the thing being asked (e.g. cellForRowAtIndexPath), or for Boolean conditions, describe the consequence of a positive value — this usually involves language like "should" (e.g. windowShouldClose, tableView:shouldSelectRowAtIndexPath:).
In your case, it's not entirely clear what "should" happen as a result of the delegate returning true or false, but here's a guess:
protocol EditorDelegate {
func editor(_ editor: Editor, shouldShowDefaultResultsForSearch query: String) -> Bool
}
The use of return flags is often considered problematic for many reasons. See this link for one of many explanations of the arguments for and against. Nevertheless, one solution that might help in this case would be to use a typealias.
typealias useDefaultBehaviour = Bool
protocol EditorDelegate {
func didStartSearch(query: String) -> useDefaultBehaviour
}
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.
I have a protocol Reusablethat has a static function static func reuseId() -> String and a protocol extension that defines the default implementation for the function. Then, I implemented a extension on UITableViewCell to conform to the Reusable protocol. I can now use the function on my TableViewCells without a problem: SomeTableViewCell.reuseId().
The Problem I have is with Generics. I have a generic class that has a generic parameter of the type UITableViewCell:
internal class SomeClass<CellType: UITableViewCell>: NSObject {
...
}
I want to be able to use the function specified in Reusable in my generic class on CellType but unfortunately this does not work as expected. The compiler always generates the error Type 'CellType' has no member 'reuseId'.
Does anybody know why this is happening? Is there a workaround?
I am using Xcode 7.0 with Swift 2.0.
Greetings from Germany
UPDATE: Here is some sample code that better shows my problem:
import UIKit
protocol Reusable {
static func reuseId() -> String
}
extension Reusable {
static func reuseId() -> String {
return String(self).componentsSeparatedByString(".").last!
}
}
extension UITableViewCell: Reusable { }
class SomeGenericClass<CellType: UITableViewCell> {
func someFunction() {
let reuseIdentifier = CellType.reuseId()
}
}
This Code will generate the above error but I do not quite understand why this happens. I think the main difference to the sample code that jtbandes posted is that I use a static function.
UPDATE: The issue is fixed in Xcode 8.3 beta 2. The sample code above now works as expected (after migrating it to Swift 3 of course).
That's an interesting problem. Your code seems like it should work; you might want to file an enhancement request.
Here's a workaround that seems to work correctly:
class SomeGenericClass<CellType: Cell> {
func someFunction() {
let reuseIdentifier = (CellType.self as Reusable.Type).reuseId()
}
}
Another (workaround) way to get what you need:
class GenericExample<CellType:UITableViewCell where CellType:Reusable>
{
func test() -> String {
return CellType.reuseId()
}
}
GenericExample<UITableViewCell>().test() // returns "UITableViewCell"
GenericExample<MyCell>().test() // returns "MyCell"
I'm trying to create a protocol in Swift I can use for object construction. The problem I'm running into is that I need to store the type information so the type can be constructed later and returned in a callback. I can't seem to find a way to store it without either crashing the compiler or creating build errors. Here's the basics (a contrived, but working example):
protocol Model {
init(values: [String])
func printValues()
}
struct Request<T:Model> {
let returnType:T.Type
let callback:T -> ()
}
We have a simple protocol that declares a init (for construction) and another func printValues() (for testing). We also define a struct we can use to store the type information and a callback to return the new type when its constructed.
Next we create a constructor:
class Constructor {
var callbacks: [Request<Model>] = []
func construct<T:Model>(type:T.Type, callback: T -> ()) {
callback(type(values: ["value1", "value2"]))
}
func queueRequest<T:Model>(request: Request<T>) {
callbacks.append(request)
}
func next() {
if let request = callbacks.first {
let model = request.returnType(values: ["value1", "value2"])
request.callback(model)
}
}
}
A couple things to note: This causes a compiler crash. It can't figure this out for some reason. The problem appears to be var callbacks: [Request<Model>] = []. If I comment out everything else, the compiler still crashes. Commenting out the var callbacks and the compiler stops crashing.
Also, the func construct works fine. But it doesn't store the type information so it's not so useful to me. I put in there for demonstration.
I found I could prevent the compiler from crashing if I remove the protocol requirement from the Request struct: struct Request<T>. In this case everything works and compiles but I still need to comment out let model = request.returnType(values: ["value1", "value2"]) in func next(). That is also causing a compiler crash.
Here's a usage example:
func construct() {
let constructor = Constructor()
let request = Request(returnType: TypeA.self) { req in req.printValues() }
//This works fine
constructor.construct(TypeA.self) { a in
a.printValues()
}
//This is what I want
constructor.queueRequest(request)
constructor.next() //The callback in the request object should be called and the values should print
}
Does anyone know how I can store type information restricted to a specific protocol to the type can later be constructed dynamically and returned in a callback?
If you want the exact same behavior of next I would suggest to do this:
class Constructor {
// store closures
var callbacks: [[String] -> ()] = []
func construct<T:Model>(type:T.Type, callback: T -> ()) {
callback(type(values: ["value1", "value2"]))
}
func queueRequest<T:Model>(request: Request<T>) {
// some code from the next function so you don't need to store the generic type itself
// **EDIT** changed closure to type [String] -> () in order to call it with different values
callbacks.append({ values in
let model = request.returnType(values: values)
request.callback(model)
})
}
func next(values: [String]) {
callbacks.first?(values)
}
}
Now you can call next with your values. Hopefully this works for you.
EDIT: Made some changes to the closure type and the next function
Unfortunately there is no way to save specific generic types in an array and dynamically call their methods because Swift is a static typed language (and Array has to have unambiguous types).
But hopefully we can express something like this in the future like so:
var callbacks: [Request<T: Model>] = []
Where T could be anything but has to conform to Model for example.
Your queueRequest method shouldn't have to know the generic type the Request it's being passed. Since callbacks is an array of Request<Model> types, the method just needs to know that the request being queued is of the type Request<Model>. It doesn't matter what the generic type is.
This code builds for me in a Playground:
class Constructor {
var callbacks: [Request<Model>] = []
func construct<T:Model>(type:T.Type, callback: T -> ()) {
callback(type(values: ["value1", "value2"]))
}
func queueRequest(request: Request<Model>) {
callbacks.append(request)
}
func next() {
if let request = callbacks.first {
let model = request.returnType(values: ["value1", "value2"])
request.callback(model)
}
}
}
So I found an answer that seems to do exactly what I want. I haven't confirmed this works yet in live code, but it does compile without any errors. Turns out, I needed to add one more level of redirection:
I create another protocol explicitly for object construction:
protocol ModelConstructor {
func constructWith(values:[String])
}
In my Request struct, I conform to this protocol:
struct Request<T:Model> : ModelConstructor {
let returnType:T.Type
let callback:T -> ()
func constructWith(values:[String]) {
let model = returnType(values: values)
callback(model)
}
}
Notice the actual construction is moved into the Request struct. Technically, the Constructor is no longer constructing, but for now I leave its name alone. I can now store the Request struct as ModelConstructor and correctly queue Requests:
class Constructor {
var callbacks: [ModelConstructor] = []
func queueRequest(request: Request<Model>) {
queueRequest(request)
}
func queueRequest(request: ModelConstructor) {
callbacks.append(request)
}
func next() {
if let request = callbacks.first {
request.constructWith(["value1", "value2"])
callbacks.removeAtIndex(0)
}
}
}
Note something special here: I can now successfully "queue" (or store in an array) Request<Model>, but I must do so indirectly by calling queueRequest(request: ModelConstructor). In this case, I'm overloading but that's not necessary. What matters here is that if I try to call callbacks.append(request) in the queueRequest(request: Request<Model>) function, the Swift compiler crashes. Apparently we need to hold the compiler's hand here a little so it can understand what exactly we want.
What I've found is that you cannot separate Type information from Type Construction. It needs to be all in the same place (in this case it's the Request struct). But so long as you keep construction coupled with the Type information, you're free to delay/store the construction until you have the information you need to actually construct the object.