> On Jun 17, 2017, at 21:12, Ben Cohen via swift-users <swift-users@swift.org>
> wrote:
>
> In order for this defaulting subscript technique to work as intended, the
> subscript { get } needs to be called, then the mutation happens, then the
> subscript { set } writes the mutated value back, including adding it for the
> first time it the default was needed.
>
> Reference types, not being value types, skip the write-back part, because
> they shouldn’t need writing back – they should just get amended in place,
> because they’re reference types.
>
> Except this particular technique is relying on it.
>
> This is probably worth a bug report, though I’m not sure if there’s an easy
> fix. The alternative is that Dictionary.subscript(_: default:) be made a
> mutating get that sets the default if not present, even without the set.
> There’s downsides to this though: you would no longer be able to use this
> defaulting subscript with immutable dictionaries, and getting a default value
> would add it which might be very unexpected.
We could say "extension Dictionary where Value: class", but that'll only fix it
when the compiler knows `Value` has reference semantics (and lead to even more
confusion when used in generic functions without the "T: class" part).
Could we check if `Value: class` within the existing setter? Or if that's what
we already do to skip the write-back, skip the skipping when one of the values
is nil?
- Dave Sweeris
_______________________________________________
swift-users mailing list
swift-users@swift.org
https://lists.swift.org/mailman/listinfo/swift-users
