[swift-evolution] Setter methods for vars

[Xiaodi Wu]

As an additive feature it's unlikely to be considered in the Swift 3
timeline anyway. You can search the list archive for previous discussion on
lenses. I think David is saying that this proposal looks like it's asking
for a special case where a more general solution might be appropriate.
On Tue, Jun 28, 2016 at 6:11 PM Michael Peternell via swift-evolution <
swift-evolution at swift.org> wrote:

> Really?? Or we just have #set and #get and no lenses, and it's done for
> Swift 3?
>
> I never heard of lenses (Google does not help here). Was this serious or
> were you joking? Unless you can explain why #set and #get without lenses
> would be bad... or maybe #set and #get *are* lenses, in which case I'm not
> sure what you were trying to say. Reflexion -> Reflection?
>
> -Michael
>
> > Am 29.06.2016 um 00:55 schrieb David Hart via swift-evolution <
> swift-evolution at swift.org>:
> >
> > This looks like lenses. I think we need to wait until after Swift 3 to
> discuss it, and come up with a bigger design that ties to reflexion.
> >
> >> On 28 Jun 2016, at 22:04, Michael Peternell via swift-evolution <
> swift-evolution at swift.org> wrote:
> >>
> >> So you're proposing that `#set(aVariableName)` should translate to
> `{aVariableName=$0}`, right? Where aVariableName can be any valid lvalue
> like `self.users` or `users` or `vc.viewControllers`..
> >>
> >> I think this would be a good extension to Swift. (`users.set` does not
> work BTW, because maybe the `users` object has a `set` property.. maybe I
> wanted to refer to the `set` property which also happens to refer to a
> closure value.)
> >>
> >> `#set(aVariableName)` also feels consistent with the
> `#keyPath(aVariableName)` property and falls into a similar category. Maybe
> `#setter(aVariableName)` would be even more clear? Furthermore, I want to
> additionally propose to introduce `#get(aVariableName)` (or
> `#getter(aVariableName)`) too.
> >>
> >> -Michael
> >>
> >>> Am 28.06.2016 um 20:18 schrieb Austin Feight via swift-evolution <
> swift-evolution at swift.org>:
> >>>
> >>> Proposal:
> >>>
> >>> I propose adding setter methods to vars, which could look something
> like this: `ApiClient().fetchUsers().then(#set(users))`
> >>>
> >>> Initially I thought it should work like this:
> `ApiClient().fetchUsers().then(users.set)`
> >>> but to accomplish a line of code that flows grammatically, I believe
> putting "set" where it would naturally fall if the code was being read as a
> sentence is more Swifty.
> >>>
> >>> Rationale:
> >>>
> >>> The following code makes me smile:
> >>>
> >>> ApiClient().fetchUsers().then(displayUsers)
> >>>
> >>> It exemplifies the beauty of Swift. First-class functions make this
> line of code read very well. Consider some alternatives:
> >>>
> >>> 1. ApiClient().fetchUsers().then { displayUsers($0) }
> >>> 2. ApiClient().fetchUsers().then { users in displayUsers(users) }
> >>> 3. ApiClient().fetchUsers().then { (users: [User]) in
> displayUsers(users) }
> >>>
> >>> Using the lessons learned from Swift API Design Guidelines (WWDC 2016
> Session 403) having an emphasis on clarity, my analysis of the alternatives
> is:
> >>>
> >>> 1. $0 adds no additional information as to the type or explanation of
> what the argument is, thus adding nothing to the line of code for clarity,
> and therefore should be omitted
> >>> 2. adding "users" also adds nothing to the clarity of the code. The
> function, properly, contains the information necessary to reason about the
> argument it takes and what it does, and therefore adding "users" is
> redundant
> >>> 3. Not only is "users" redundant, but also is the explicit type label.
> The `displayUsers` method will only accept one type of argument, so we're
> duplicating information that the compiler (and autocomplete) already gives
> us
> >>>
> >>> With this I conclude that
> `ApiClient().fetchUsers().then(displayUsers)` is the Swiftiest option.
> >>> I want to extend this same logic to when I find myself writing code
> like this:
> >>>
> >>> ApiClient().fetchUsers().then { users in
> >>> self.users = users
> >>> }
> >>>
> >>> or alternatively, because "users" is likely redundant information
> again,
> >>>
> >>> ApiClient().fetchUsers().then { self.users = $0 }
> >>>
> >>> Personally I steer clear of `$0` as much as possible, because I very
> rarely feel that it provides the information necessary for code clarity.
> But beyond that, this code no longer reads as nicely as the code we had
> before.
> >>>
> >>> Whereas `ApiClient().fetchUsers().then(displayUsers)` flows nicely as
> a sentence and reads grammatically, `ApiClient().fetchUsers().then {
> self.users = $0 }` no longer does.
> >>>
> >>> I think this feature could have a simple implementation where the
> compiler replaces `#set(X)` with `{ X = $0 }`, and I believe it would go a
> long way with respect to code clarity, especially when X is something
> longer like `self.view.bounds.origin.x`
> >>>
> >>>
> >>> Looking forward to hearing thoughts from the community,
> >>> Austin Feight
> >>> _______________________________________________
> >>> swift-evolution mailing list
> >>> swift-evolution at swift.org
> >>> https://lists.swift.org/mailman/listinfo/swift-evolution
> >>
> >> _______________________________________________
> >> swift-evolution mailing list
> >> swift-evolution at swift.org
> >> https://lists.swift.org/mailman/listinfo/swift-evolution
> >
> > _______________________________________________
> > swift-evolution mailing list
> > swift-evolution at swift.org
> > https://lists.swift.org/mailman/listinfo/swift-evolution
>
> _______________________________________________
> swift-evolution mailing list
> swift-evolution at swift.org
> https://lists.swift.org/mailman/listinfo/swift-evolution
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-evolution/attachments/20160628/b048019d/attachment.html>