This is correct behavior: Swift does not infer parameter or return types from the bodies of multi-statement closures. But the diagnostic could be a lot better.

@belkadan OK, Swift does not infer parameter or return types from the bodies of multi-statement closures. This is the current behavior.

But is it the correct behavior? Let's say that this the topic of the issue, if you prefer.

The user who initially complained (groue/GRDB.swift#53 writes code like:

f {
    let x = 1
    return x
}

Is it normal to punish him for not writing a single-line closure?

I think not.

It's a technical issue as much as a design issue. I talked a bit about it on swift-evolution back in December; see the last section in https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151214/002583.html.

But the diagnostic could be a lot better.

Should the behavior remain, a fix-it would indeed be very much welcome.

Please take in account that the burden of this behavior weights on the shoulders of library developers: users get stuck, find no way to fix their trouble, and eventually report an issue. Answering issues takes time.

Thanks for the link to the rationale, Jordan

Note to future readers: Observe that this issue does not arise if there's something else (e.g. another input) that resolves the generic. This compiles:

func f<T>(_ i:T, _ ff:(T) -> T) -> T {
    return ff(i)
}
let r = f(2) {
    let result = $0*10
    return result
}

It is only when all the compiler has to go on is the return type that the issue arises.

Fixed with SE-0326