[SR-7008] Subclass existential doesn't satisfy generic class constraints

[karwa]

Previous ID	SR-7008
Radar	None
Original Reporter	@karwa
Type	Bug
Status	Resolved
Resolution	Duplicate

Environment

Xcode 9.2

Apple Swift version 4.1 (swiftlang-902.0.34 clang-902.0.30)

Additional Detail from JIRA

Votes	0
Component/s	Compiler
Labels	Bug
Assignee	None
Priority	Medium

md5: 10ba50906ce8e750485046f8da2fb431

duplicates:

• SR-55 non-@objc protocol existentials do not conform to their own protocol type

Issue Description:

Example:

class AClass {}
protocol AProto { var value: Int { get } }

class Box<T> { 
  private var _boxed: T
  init(_ b: T) { _boxed = b } 
}
extension Box where T: AClass { 
  var boxed: T { 
    get { return _boxed }
    set { _boxed = newValue } 
  } 
}

class One: AClass, AProto { var value: Int { return 1 } }
class Two: AClass, AProto { var value: Int { return 2 } }

let a: Box<AClass & AProto> = Box(One())
print(a.boxed) // error: 'AClass & AProto' is not convertible to 'AnyObject'
a.boxed = Two()
print(a.boxed)

let b: AClass & AProto = One()
print(b is AClass) // 'true'

See discussion at: https://forums.swift.org/t/confused-by-generic-behaviour-not-clear-if-it-is-a-bug/9823

[swift-ci]

Comment by Marc Palmer (JIRA)

I completely understand how Swift experts may say "you're holding it wrong" but I think there is a much wider and more important point here that this doesn't look wrong at all and hence any APIs built like this will be very hard for developers to reason about.

I am still struggling to internalise this nuance with existentials, and I've developed in many languages (C, C++, Object Pascal, ObjC, JS, 68000 ASM) for over 30 years!