[swift-evolution] two protocols with the same method name

Howard Lovatt howard.lovatt at gmail.com
Fri Jan 8 23:25:51 CST 2016


Thanks, I understand what you are driving at now. 

This is true in many languages that have interfaces (protocols), e.g. Java and duck-typed languages, it doesn’t seem to be a big deal. Languages that have traits or mixins, e.g. Scala, can deal with this.

I suggest that it isn’t a big enough problem to worry about. It might go away if protocols morph into traits :).

> On 9 Jan 2016, at 2:09 PM, Brent Royal-Gordon <brent at architechies.com> wrote:
> 
>> I don’t really get what you are driving at.
> 
> The point is that, although `A` and `B` both require properties with the same name, they expect different semantics from that property. Let's maybe give these more concrete names so you can understand the idea:
> 
> 	protocol Marriageable {
> 		var ring: String? { get set }	// File name of image of this person's wedding ring.
> 	}
> 	protocol CallReceivable {
> 		var ring: String? { get set } 	// File name of ringtone to be used for this person.
> 	}
> 
> 	struct Person: Marriageable, CallReceivable {
> 		var ring: String?
> 	}
> 
> Of course a person is marriageable and can this have "a ring", and of course you can also receive a call from them and they can thus have "a ring". But in reality, the "ring" that satisfies one of these things will not work for the other. If your best friend gets married and you add an image of the ring, then the next time your friend calls you, the phone ringing screen will try to play a JPEG as an MP3.
> 
> The "ring" example is, of course, slightly contrived, but I'm sure you can imagine a similar problem with real names, where you end up using the same term for two different and incompatible things.
> 
> What the OP is basically asking is, when Swift sees the same type conforming to Marriageable and CallReceivable, should it optimistically assume that the `ring` properties they both require are compatible and allow the code to pass through without comment? Or should it pessimistically assume that the `ring` properties are incompatible and emit a warning or error about them?
> 
> -- 
> Brent Royal-Gordon
> Architechies
> 



More information about the swift-evolution mailing list