Why and how any ErrorType can always be casted to NSError?

Question

Eonil OP

Created Sep ’15

Replies 4

Boosts 0

Views 4.4k

Participants 5

I installed Xcode 7 GM (7A218) today, and I got an interesting warning.

Dialogue.swift:91:25: Conditional cast from 'T' to 'NSError' always succeeds

Here's my source code.

  static func runErrorAlertModally<T: ErrorType>(error: T) {
       func makeAlert() -> NSAlert {
            if let error = error as? NSError { // Conditional cast from 'T' to 'NSError' always succeeds
                 return NSAlert(error: error)
            }
            else {
                 let a = NSAlert()
                 a.messageText = "\(error)"
                 return a
            }
       }
       makeAlert().runModal()
  }

AFAIK, ErrorType is just a marker type to mark a type can be thrown. But "why" and "how" `ErrorType` can be casted into `NSError` without any special treatment?

Boost

Answer 1

Tia Lapis OP

Sep ’15

Because ErrorType has NSError as implicit Superclass.

0

Answer 2

Vision Pro Engineer OP

Apple

Sep ’15

That's incorrect, ErrorType is a protocol. The reason this works is because of "compiler magic." The compiler automatically emits the code necessary to translate between any ErrorType and NSError.

0

Answer 3

Tia Lapis OP

Sep ’15

Ah my NSError 😉 - I assumed the compilermagic was a implicit superclass in the background.

0

Answer 4

Wallacy OP

Sep ’15

Probably he only said with compiler does not use a class to do this and the ErrorType requirements match exactly as NSError so is "Toll free bridged".

But magic is more fun.

0