[swift-evolution] Ownership on protocol property requirements
matthew at anandabits.com
Sun May 7 15:02:33 CDT 2017
Sent from my iPad
> On May 7, 2017, at 1:12 PM, Xiaodi Wu via swift-evolution <swift-evolution at swift.org> wrote:
> Today these keywords have no meaning inside a protocol, so clearly it should be an error to use it in that context. I agree with Jordan that the error should be on the protocol.
> It's entirely a different conversation whether the keyword should have meaning or not. If it should, it seems to me it should be limited to protocols that are limited to classes. But that's an additive feature we can discuss later.
Why would it make sense to limit this to class-constrained protocols? It would obviously make sense to limit it to properties of class or class-constrained type but I see no reason why an arbitrary restriction to class-constrained protocols would make sense even if that was how it is most commonly used.
> The source-breaking bug fix that is more pressing today is removing meaningless keywords that can be misleading to users, because they have no effect but look like they should.
>> On Sun, May 7, 2017 at 11:00 Goffredo Marocchi via swift-evolution <swift-evolution at swift.org> wrote:
>> It would be useful to have a longer discussion on this as... I think that weak has a place there and should be enforced as a protocol is the public facing interface/api for the types who decide to adopt it.
>> Sent from my iPhone
>> > On 7 May 2017, at 15:41, Adrian Zubarev via swift-evolution <swift-evolution at swift.org> wrote:
>> > browse
>> swift-evolution mailing list
>> swift-evolution at swift.org
> swift-evolution mailing list
> swift-evolution at swift.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the swift-evolution