Fix possible ambiguous use of when(fulfilled:) in tests

[RomanPodymov]

Hello.
Thank you for PromiseKit.
I found out that in testUnhandledErrorHandlerDoesNotFireForStragglers and testAllSealedRejectedFirstOneRejects it is hard to determine which version of when(fulfilled:) is being used. As for me, there should be the ambiguous use of when(fulfilled:) error. My suggestion is to store when(fulfilled: p1, p2, p3)'s result in a variable, it helps to know which version of when(fulfilled:) is being used.

[RomanPodymov]

There can be also let whenFulfilledP1P2P3: Promise<Void> = when(fulfilled: p1, p2, p3).

[RomanPodymov]

There can be also let whenFulfilledP1P2P3: Promise<(Void, Void, Void)> = when(fulfilled: p1, p2, p3).

[mxcl]

Forgive me, but I don’t see how this is ambiguous, ambiguous with which other version of when?

[RomanPodymov]

Hello @mxcl
For instance, in

let p1 = Promise<Void>(error: Error.test)
let p2 = after(.milliseconds(100)).done { throw Error.straggler }
let p3 = after(.milliseconds(200)).done { throw Error.straggler }

when(fulfilled: p1, p2, p3).catch { error -> Void in
    XCTAssertTrue(Error.test == error as? Error)
    ex1.fulfill()
}

As I can see in the debugger func when<U: Thenable, V: Thenable, W: Thenable>(fulfilled pu: U, _ pv: V, _ pw: W) -> Promise<(U.T, V.T, W.T)> is using here. But public func when<U: Thenable>(fulfilled promises: U...) -> Promise<Void> where U.T == Void is suitable here as well:

(when(fulfilled: p1, p2, p3) as Promise<Void>).catch { error -> Void in
    XCTAssertTrue(Error.test == error as? Error)
    ex1.fulfill()
}

Maybe there is a rule why func when<U: Thenable, V: Thenable, W: Thenable>(fulfilled pu: U, _ pv: V, _ pw: W) -> Promise<(U.T, V.T, W.T)> has a higher priority.

[mxcl]

Oh ok. I’m not sure it matters per se, but why not.