类继承和运算符重载

Question

我正在通过一个关于SFML的小项目学习C++，我想扩展sf::Vector2<T>类，它表示一个2D数学向量。这个类的声明是available here。特别是，我想在类中添加一些方法，用于范数计算、旋转等。

这就是我到目前为止所知道的：

#include <cmath>
#include <SFML/System/Vector2.hpp>
#include <SFML/Graphics/Transform.hpp>
#include <ostream>

namespace dw::math {

template <typename T>
class Vector2 : public sf::Vector2<T> {
public:
    // Don't really know what does that exactly means, explainations are welcome !
    using sf::Vector2<T>::Vector2;

    template <typename U>
    explicit Vector2(const U &vector) : sf::Vector2<T>(vector) {}

    // This ctor is for implicit conversion from base, don't know if it's really useful ?
    Vector2(sf::Vector2<T> &vector) : sf::Vector2<T>(vector) {}

    // Some methods here ...

    friend std::ostream & operator<<(std::ostream &os, const math::Vector2<T>& right) {
        os << '{' << right.x << ", " <<right.y << '}';
        return os;
    }
};

在后面的代码中，我声明了这样一个结构：

struct Derivative {
    dw::math::Vector2f dPos;
    dw::math::Vector2f dVel;

    Derivative() = default;
    Derivative(const dw::math::Vector2f &dPos, const dw::math::Vector2f &dVel) : dPos(dPos), dVel(dVel) {}

    Derivative operator+(const Derivative &rhs) const {
        return Derivative(
            dPos + rhs.dPos,
            dVel + rhs.dVel
        );
    }
}

Derivative的operator+重载的返回部分无法编译：

/home/palra/Documents/Projets/dw/src/physics/World.cpp: In member function ‘Derivative Derivative::operator+(const Derivative&) const’:
/home/palra/Documents/Projets/dw/src/physics/World.cpp:24:18: error: invalid user-defined conversion from ‘sf::Vector2<float>’ to ‘const Vector2f& {aka const dw::math::Vector2<float>&}’ [-fpermissive]
             dPos + rhs.dPos,
             ~~~~~^~~~~~~~~~
In file included from /home/palra/Documents/Projets/dw/src/physics/Body.h:8:0,
                 from /home/palra/Documents/Projets/dw/src/physics/World.h:10,
                 from /home/palra/Documents/Projets/dw/src/physics/World.cpp:5:
/home/palra/Documents/Projets/dw/src/physics/../math/Vector2.h:30:5: note: candidate is: dw::math::Vector2<T, <template-parameter-1-2> >::Vector2(sf::Vector2<T>&) [with T = float; <template-parameter-1-2> = void] <near match>
     Vector2(sf::Vector2<T> &vector) : sf::Vector2<T>(vector) {}

我不明白这为什么行不通。表达式dPos + rhs.dPos与sf::Vector2<float> operator +(const sf::Vector2<float>& left, const sf::Vector2<float>& right)的调用是兼容的，因为dw::math::Vector2<float>通过继承就是sf::Vector2<float>。然后，这个表达式将产生一个sf::Vector2<float>，我想这要归功于非显式构造函数，它可以赋值给dw::math::Vector2f。我哪里错了？我该怎么让它工作呢？

Answer 1

sf::Vector<T>不是设计用来作为派生类使用的，因为这样一种更好的方法是编写免费函数来扩展sf::Vector<T>的功能，这对于操作员来说是无缝工作的。不幸的是，对于普通的函数调用，C++还不像C#那样支持扩展。