Voids

I recently ran into a problem where I was confused by the difference between () and Void in Swift.

TL;DR Void is a type, but not a value, and () is both a type and a value.

Sometimes they’re the same

My confusion arose from the fact that I think of them as interchangeable, when it comes to function signatures. For example, these two are the same:

func asyncWork1(completion: @escaping (String) -> Void)
func asyncWork2(completion: @escaping (String) -> ())

But not always

However, given this generic enum definition¹:

enum Result<T> {
    case success(T)
    case failure(Error)
}

I was trying to represent the success case, but without a value. So I tried this:

let result1: Result<Void> = .success(Void)

But that gave this (somewhat cryptic) error:

error: member 'success' in 'Result<Void>' (aka 'Result<()>') produces result of type 'Result<T>', but context expects 'Result<Void>' (aka 'Result<()>')

It’s because Void is just a type, not a value, and the compiler is expecting a value here.

So, to fix it, we use () instead, which is a value.

let result2: Result<Void> = .success(())

And, since we’ve realized that the issue is Void is a type, not a value, it seems obvious to try this, which also works (but does feel a bit odd to me ²):

let result3: Result<Void> = .success(Void())

Parting Thoughts

I do wonder a bit about why () is allowed as a type as well as a value ³. I think it has to do with the subtle details of Swift’s syntax, and tuples, and the desire to be able to write () -> () as a function signature, instead of Void -> Void? But, it did lead to some confusion in this case (no pun intended).

Thanks, as usual, to Jacob for pointing out the type/value difference to me here.