对于通用标题很抱歉,如果没有示例,很难描述问题.
假设我定义了以下限制为Equatable类型的泛型函数:
func test<T: Equatable>(expect expected: T,run: () -> T) { let value = run() if value == expected { print("OK") } else { print("Expected: \(expected),Actual: \(value)") } }
这是使用所述函数的示例:
test(expect: 100) { 10 * 10 } // prints "OK" test(expect: 1000) { 10 * 10 } // prints "Expected: 1000,Actual: 100"
当然,我可以存储值而不是使用文字:
let e = 100 test(expect: e) { e } // prints "OK"
到目前为止一切都那么好,一切都按预期工作(没有双关语意).
现在让我们用一个数组来尝试:
test(expect: [1,2]) { [1,2] } // prints "OK"
事情再次成功.
但现在我们尝试这个:
let a = [1,2] test(expect: a) { a } // error: cannot convert value of type '() -> [Int]' to expected argument type '() -> _'
所以我一直在努力的问题是:为什么这不起作用?
游乐场正确地推断出a的类型为[Int],所以() – >的期望值在哪里? _ 来自?
尝试上一个例子的一堆变体:
test(expect: a) { return a } test(expect: a) { return a as [Int] } test(expect: a as [Int]) { return a as [Int] } test(expect: [1,2]) { a } test(expect: [1,2] as [Int]) { a }
它们都会导致同样的问题.出于某种原因,Swift似乎认为函数期望() – > _.
所以也许这只是因为数组不是Equatable,但这可行:
let a = [1,2] [1,2] == [1,2] a == a
我以为我非常了解泛型,我完全被这个难过了.这是Swift中的错误还是我对test()的定义中的错误?甚至可以实现目标吗?
解决方案
感谢@ Sulthan在下面的回答,我能够编写此函数的另一个版本来处理数组大小写(以及任何SequenceType):
public func test<T: SequenceType where T.Generator.Element: Equatable>(expect expected: T,run: () -> T) { let result = run() // Note: zip() will stop at the shorter array,so this implementation isn't correct,don't use it (it will incorrectly end up saying [1] == [1,2]). This code is just here to demonstrate the function's generic constraint. let eq = zip(expected,result).filter(!=).isEmpty if eq { print("OK") } else { print("Expected: \(expected),Actual: \(result)") } } let a: [Int] = [1,2] test(expect: [1,2]) { a } // prints "OK" test(expect: [1,3]) { a } // prints "Expected: [1,3],Actual: [1,2]"