Swift BinaryFloatingPoint 泛型在 log10() 调用上暴露编译错误

Question

T 应该遵守什么协议(protocol)来传递这个错误？

class func magnitude<T: BinaryFloatingPoint>(_ n: T) where T: CVarArg {
    // …

    let p = Int(log10(n))

    // …
}

错误如下:

Cannot invoke 'log10' with an argument list of type '(T)'

Answer 1

根据 Martin R 的技巧，我想出了以下内容(我添加了虚构的代码，以便可以编译它)，效果很好:

protocol FloatingPointMathType : BinaryFloatingPoint, CVarArg {
    var _log10Value : Self { get }
}

extension Float: FloatingPointMathType {
    var _log10Value : Float {return log10(self)}
}

extension Double: FloatingPointMathType {
    var _log10Value : Double {return log10(self)}
}

extension CGFloat: FloatingPointMathType {
    var _log10Value : CGFloat {return log10(self)}
}

func log10<T:FloatingPointMathType>(_ x:T) -> T {return x._log10Value}

class Format {
    class func magnitude<T: BinaryFloatingPoint>(_ n: T) -> Int {

        let p = Int(log10(n))

        return p
    }
}

let d: Double = 12.0
print(Format.magnitude(d))