0

考虑以下 C++ 代码:

#include <iostream>

using std::cout;

class A
{
public:
    int a;

    A():a(0)
    {
        cout << "A constructor\n";
    }

    virtual void f()
    {
        cout << "f inside A\n";
    }
};

class C : public A
{
public:
    int c;

    virtual void f()
    {
        cout << "f inside C\n";
    }

    C():c(0)
    {
        cout << "C constructor\n";
    }
};

int main()
{
    A varA = C();

    cout << "Size of C class: " << sizeof(C) << "\n";
    cout << "Size of varA object: " << sizeof(varA) << "\n";

    C* varC = static_cast<C*>(&varA);
    varC->f();

    cout << "varC->a is " << varC->a << "\n";
    cout << "varC->c is " << varC->c << "\n";
}

这个程序的输出是:

A constructor
C constructor
Size of C class: 16
Size of varA object: 8
f inside A
varC->a is 0
varC->c is 1726166356

我用 class 的构造函数初始化varA对象C。调用了 A 和 C 类的构造函数,但它们varA只是一个A对象。我将地址varA转换为C*类型并尝试调用它的f()函数,但是它打印了f()class 的函数A,所以我推断它是使用早期绑定机制来调用它。我想如果我调用派生类的构造函数,就像这种情况,如果我调用了基构造函数,我会获得相同的对象。我认为唯一的区别是调用了其他构造函数。我的假设是正确的还是有任何其他差异?

4

2 回答 2

4

切片的经典示例。A varA = C();为您留下静态和动态类型的对象A。结果,C* varC = static_cast<C*>(&varA);表现出未定义的行为。

于 2016-12-14T14:51:37.907 回答
1

您可以将完整的派生类存储在基类指针中,但是:

int main() {        
  A* varA = new C();
  C* varC = static_cast<C*>(varA);
  varC->f();

  cout << "varC->a is " << varC->a << endl;
  cout << "varC->b is " << varC->b << endl;
  cout << "varC->c is " << varC->c << endl;
} // oops, forgot to delete varA/varC, memory leak!
于 2016-12-14T15:29:55.910 回答