开发者

Operator overloading outside class [duplicate]

开发者 https://www.devze.com 2022-12-22 10:47 出处:网络
This question already has answers here: 开发者_C百科 What are the basic rules and idioms for operator overloading?
This question already has answers here: 开发者_C百科 What are the basic rules and idioms for operator overloading? (8 answers) Closed 5 years ago.

There are two ways to overload operators for a C++ class:

Inside class

class Vector2
{
public:
    float x, y ;

    Vector2 operator+( const Vector2 & other )
    {
        Vector2 ans ;
        ans.x = x + other.x ;
        ans.y = y + other.y ;
        return ans ;
    }
} ;

Outside class

class Vector2
{
public:
    float x, y ;
} ;

Vector2 operator+( const Vector2& v1, const Vector2& v2 )
{
    Vector2 ans ;
    ans.x = v1.x + v2.x ;
    ans.y = v1.y + v2.y ;
    return ans ;
}

(Apparently in C# you can only use the "outside class" method.)

In C++, which way is more correct? Which is preferable?


The basic question is "Do you want conversions to be performed on the left-hand side parameter of an operator?". If yes, use a free function. If no, use a class member.

For example, for operator+() for strings, we want conversions to be performed so we can say things like:

string a = "bar";
string b = "foo" + a;

where a conversion is performed to turn the char * "foo" into an std::string. So, we make operator+() for strings into a free function.


First: the two different ways are really "overload as a member" and "overload as a non-member", and the latter has two different ways to write it (as-friend-inside class definition and outside class definition). Calling them "inside class" and "outside class" is going to confuse you.


Overloads for +=, +, -=, -, etc. have a special pattern:

struct Vector2 {
  float x, y;
  Vector2& operator+=(Vector2 const& other) {
    x += other.x;
    y += other.y;
    return *this;
  }
  Vector2& operator-=(Vector2 const& other) {
    x -= other.x;
    y -= other.y;
    return *this;
  }
};
Vector2 operator+(Vector2 a, Vector2 const& b) {
  // note 'a' is passed by value and thus copied
  a += b;
  return a;
}
Vector2 operator-(Vector2 a, Vector2 const& b) { return a -= b; } // compact

This pattern allows the conversions mentioned in the other answers for the LHS argument while simplifying the implementation considerably. (Either member or non-member allows conversions for the RHS when it's passed either as a const& or by value, as it should be.) Of course, this only applies when you do actually want to overload both += and +, -= and -, etc., but that is still common.


Additionally, you sometimes want to declare your non-member op+, etc. as friends within the class definition using the Barton-Nackman trick, because due to quirks of templates and overloading, it may not be found otherwise.


There is an excellent discussion of this issue in Meyer's Effective C++: Item 24 is "Declare non-member functions when type conversions should apply to all parameters" and Item 46 is "Define non-member functions inside templates when type conversions are desired".

0

精彩评论

暂无评论...
验证码 换一张
取 消