C ++重载运算符并返回相同的对象

Question

I'm coding a class for Vector and Matrices and I'd like to understand how can I avoid overheads and leaks when I want to overload common operators such as +, -, * and /.

For example:

int main()
{
    Vector3 aVector; //This has an address #1
    Vector3 bVector; //This has another address #2

    //rVector has an address #3
    Vector3 rVector = aVector - bVector; //What will happen here?
}

And the vector class:

class Vector3
{
    public:
        float vX, vY, vZ;
        Vector3& operator-(const Vector3& vector3)
        {
            //I want to calculate this vector with the "vector3" param
            //But then what do I return?

            //Test 1:
            Vector3 result; //This has an address #4
            result.vX = vX - vector3.vX;
            result.vY = vY - vector3.vY;
            result.vZ = vZ - vector3.vZ;

            return result; //Did I just overwrite address #3?

            //Test 2:
            vX = vX - vector3.vX;
            vY = vY - vector3.vY;
            vZ = vZ - vector3.vZ;

            return (*this); //What happened to address #3? And I just changed this vector's values and I need then again later
        }
}

What's the best way to do this?

edit : One more question, if I want to do this:

Vector3 myVector = someVector - Vector3(x, y, z);

How do I code the constructor so it doesn't do anything... bad? I'm thinking it'll be creating a new class but I won't have any means to reference it after it's used in the sentence above, can this lead to problems later?

Answer 1

Your implementation doesn't have any leaks. Every vector you create is destructed when getting out of the scope of the function.

As for the return value, I suggest you return it by value (just remove the &).

For your second question: It's not a new so there is no leak. Also, you don't need to reference it so there is no real issue.

Answer 2

Create operator- as a namespace scope function (not as a member function):

Vector3 operator-(const Vector3 &lhs, const Vector3 &rhs)
{
  Vector3 result{lhs};

  // Subtract out rhs
  ...

  // Do the math
  return result;
}

Nothing gets leaked in any case, because you haven't allocated anything off of the heap (free store).

Answer 3

Test 1 will store the answer in a new vector (result) and hypothetically return it as a reference to rVector.

rVector = result

aVector = //its original value

Test 2 will store the answer in the vector the function was called on (aVector) and returns that as a reference to rVector.

rVector = aVector

aVector = //the answer

Although using test 2 will save memory the answer is stored in two places outside the function (rVector and aVector) which may not be what you want. Test 1 is the safer option but make sure to return it by value to avoid out-of-scope errors as others here have pointed out.