Class Constructor arguments C++

问题:

I am creating a pair1 class for a x and y Cartesian coordinate system. x and y are doubles. I need to have 3 constructors.

1. No arguments, defaults x and y to zero.
2. One arguement assigns x and defaults y to zero.
3. One arugeument defaults x to zero and assigns y. I'm not sure if I am setting up the class right. I get the follwing error: pair1::pair1(double) and pair1::pair1(double) cannot be overloaded.

My class:

class pair1
{
private:
    double x;
    double y;

public:
    pair1(){ x = 0.0, y = 0.0; }    
    pair1( double a ){ x = a; y =0.0; }
    pair1(double b){ x = 0.0;  y = b; }
};

回答1:

1) no arguments, defaults x and y to zero.

That's easy

2) one arguement assigns x and defaults y to zero.

3) one arugeument defaults x to zero and assigns y.

That's a problem. How do you know, when you only have one parameter, which of the two is meant to be called? That's why you get a compilation error.

Instead - use the default constructor (the one with no parameters), full constructor (the one with both), if needed, and SetX() and SetY() to set the X and Y separately, and make distinction by the name of the function.

class pair1
{
    private:
    double x;
    double y;

    public:
    pair1( double a=0.0, double b=0.0 ){ x = a; y =b; };
                     // default value 0.0 allows to only
                     // set x, and leave y to be the default,
                     // or leave them both default.
    void SetX(double a) { x=a;};
    void SetY(double b) { y=b;};
};

回答2:

The problem is that the compiler has no way to distinguish

pair1(double a)

and

pair1(double b)

Indeed, they are the same thing except for the name of the parameter. For example:

pair1 myPair1(123.456); // Which constructor should be called?

This is called ambiguous overloading.

回答3:

pair1( double a ){ x = a; y =0.0; }
pair1(double b){ x = 0.0;  y = b; }

These are exactly same constructor. Different parameter name doesn't make any difference. All that matters for overloading is, the type(s) and number of types, and their ordering.

回答4:

I'm not sure that having default arguments except for the (0,0) case is of any use, but something like this could work:

struct X
{
    double value;
    explicit X(double v) : value(v) {}
};

struct Y
{
    double value;
    explicit Y(double v) : value(v) {}
};

class pair1
{
    private:
        double x;
        double y;

    public:
        pair1() : x(0.0), y(0.0) {}
        explicit pair1(X a) : x(a.value), y(0.0) {}
        explicit pair1(Y b) : x(0.0), y(b.value) {}
        pair1(X a, Y b) : x(a.value), y(b.value) {}  // For completeness
        pair1(Y b, X a) : x(a.value), y(b.value) {}  // For super-completeness
        pair1(double a, double b) : x(a), y(b) {}
};

Use:

pair1 aPair(X(2.0));                 // default y
pair1 anotherPair(Y(4.875));         // default x
pair1 aThirdPair(X(1.0), Y(1.0));
pair1 aForthPair(Y(100.0), X(1.0));  // keyword arguments ;-)
pair1 quintus(23.0, 45.6);