> On Feb 14, 2017, at 3:43 AM, Brent Royal-Gordon <[email protected]> > wrote: > >> On Feb 13, 2017, at 7:45 AM, Matthew Johnson <[email protected]> wrote: >> >> If you look closely, when most people say “closed enum” they mean a fixed, >> complete set of cases that are all public. But when people say “closed >> protocol” they don’t actually mean a fixed, complete set of conformances >> that are all public. They simply mean clients cannot add conformances. >> This is the semantic contract of resilient enums, not closed enums. > > Yes, our traditional terminology here has been a little bit confused. > >>> What I instead suggest is that we think of a closed enum as being like a >>> fragile (non-resilient) struct. In both cases, you are committing to a >>> particular design for the type. So I think we should give them both the >>> same keyword—something like: >>> >>> @fixed struct Person { >>> var name: String >>> var birthDate: Date >>> } >>> @fixed enum Edge { >>> case start >>> case end >>> } >>> >> >> You omitted public here. Does that mean you intend for `@fixed` to imply >> public visibility? If so, I could get behind this. But I am curious why >> you made it an attribute rather than a keyword. > > No, I'm sorry, I meant to say `@fixed public struct` and `@fixed public > enum`. I don't think `@fixed` implies public-ness, either, so it would need > to be paired with a `public` keyword. There *may* be keywords we could use > that would, like `exposed`
I agree that `fixed` (and `closed`) don’t imply `public` in terms of the colloquial meaning of the words and there is a reasonable case that `open` does. I’m not sure I like `exposed`, but maybe it’s possible to find a keyword that would more directly imply `public`. > , but I'm not sure we want to make this feature so prominent, I have some trouble getting on board with requiring an annotation *in addition* to `public` for the reasons I have already stated, and which led to `open` becoming an access modifier rather than an annotation. It’s possible I could be convinced otherwise, but I think it would require data showing that this really is a rare edge case. If the relatively frequency of closed vs resilient enums is reasonably similar to the relative frequency of public vs open enums I think there is a strong case to make them carry the same syntactic weight, as we did with `open`. > and I'm not sure how that would work with classes you want to both expose and > permit subclassing of. (Would that be `exposed open class Foo`?) Can you elaborate on what you mean by "classes you want to both expose and permit subclassing of”? Do you mean commit to the set of fields being fixed like you indicated with a struct? If so, I’m not sure that is a valuable combination and my instinct is to ban it. If we did want to support something like that it points to keeping `closed` (as in cases, subclasses and conformances) orthogonal to `fixed` (as in the set of stored properties). > >>> I don't see it mentioned here (maybe I just missed it), but even though we >>> *could* do exhaustiveness checking on non-open protocols, I'm not convinced >>> that's a good idea. Usually when you have several types conforming to a >>> protocol, you should access type-specific behavior through polymorphism, >>> not by switching on the protocol. A protocol is supposed to represent a >>> behavior, not just mark a type in some arbitrary way. >> >> I agree that you should usually be adding polymorphism, but preventing >> exhaustive switch on what is effectively a style argument seems like an >> unnecessary restriction to me. There will be times when it could be used to >> good effect. I think the community has done a pretty good job of figuring >> out how to use Swift’s many features well and don’t believe it would be >> frequently abused. > > I agree we shouldn't change the language to *prevent* bad style. But this > would go beyond that—we'd be putting specific engineering effort solely into > *enabling* bad style. At minimum, this should fall so far down our to-do list > that we'll probably never get to it. This assumes that switching over conforming types is bad style. One of the biggest problems with switching over subclasses or conforming types is the fact that you don’t get compiler verification of exhaustiveness. If the language supports exhaustive switching for closed classes and protocols this becomes a non-issue. I don’t know of any languages that support a kind of type which supports generic and dynamic dispatch as well as exhaustive switch. It may be interesting to have the ability to organize some methods by type (i.e. protocol requirements) and other methods by function (i.e. a protocol extension method with an exhaustive switch). When you have exhaustive switch these are really just two different ways to organize code. Neither one is inherently better. Each has strengths different strengths. Why not allow the language to support both and let programmers decide which organization of their code is best in a particular case? > >>> I still support this general approach. One spelling could simply be >>> `@nonopen`. Although if we don't use `closed`, we could simply use >>> `@closed` like I suggested—here it really *would* be an antonym to `open`. >> >> I like the idea of using `@nonopen` for the transitional attribute. Both >> because it “removes the openness” that `public protocol` currently implies. >> In that sense it is probably the most accurate term we could find and it’s >> also pretty concise. > > It also sounds a little bit awkward, which is normally a reason not to use > it, but perhaps that's actually a good thing in a temporary, transitional > keyword. > >>>> A similar mult-release strategy would work for migrating public enums. >>> >>> What is it that needs migrating here? Lack of exhaustiveness checking? It >>> sounds like we were planning to break that anyway in some fashion. >> >> Public enums are not currently resilient. Clients are allowed to switch >> over them without a `default` clause. This means that client code will fail >> to compile in a version of Swift where `public enum` has the resilient >> contract unless the library changes to adopt closed semantics or the client >> adds a default case. > > My thinking was that, since most existing `public` enums should probably not > be `@fixed`, we should just change the behavior and let some switch > statements break. Most `public` protocols, on the other hand, ought to > become `open`, so we should flag that change and require an explicit marker > like `@nonopen` if you really don't want to change over. But I could be > convinced otherwise. I think this hits on the basis of our disagreement. Is it really the case that *most* existing `public` enums should probably not be `@fixed`? Have you done an analysis to support this? Not just of the standard library, but also Apple’s frameworks and open source modules on Github? We might learn something interesting by doing an analysis like this. It certainly wouldn’t hurt. If it turns out that closed / fixed is *often* (say 30-40% of the cases or more) the desirable contract using an annotation would result in noisy boilerplate. On the other hand, if it turns out to be relatively rare (say 5-10%) it would become easier to forget the annotation, making an error of omission. Of course that error can be corrected pretty easily so maybe we it wouldn’t be a big deal if it really is rare. > > -- > Brent Royal-Gordon > Architechies > _______________________________________________ swift-evolution mailing list [email protected] https://lists.swift.org/mailman/listinfo/swift-evolution
