在下面的代码中:

智能指针数据成员pImpl(class Impl)和原始指针pc(class CAT)都是不完整的数据类型，Widget.h中没有这两个类的定义

//控件.h

#ifndef W_H_
#define W_H_
#include <memory>

class Widget { 
    public:
        Widget();
        ~Widget() {    //
            delete pc; // I know I should put ~Widget to .cpp
                       // I just want to show the difference in behavior
                       // between raw pointer and smart pointer(both has incomplete type)
                       // when widget object destructs 
        }
    private:
        struct Impl; 
        std::shared_ptr<Impl> pImpl;  // use smart pointer
        struct CAT;
        CAT *pc;  //raw pointer
};
#endif

//控件.cpp

#include "widget.h"

#include <string>
#include <vector>
#include <iostream>
using namespace std;
struct Widget::CAT 
{
    std::string name;
    CAT(){cout<<"CAT"<<endl;}
    ~CAT(){cout<<"~CAT"<<endl;}
};      


struct Widget::Impl {
    std::string name;
    Impl(){cout<<"Impl"<<endl;}
    ~Impl(){cout<<"~Impl"<<endl;}
};


Widget::Widget()  
    : pImpl(std::make_shared<Impl>()),
      pc(new CAT)
{} 

// main.cpp

#include "widget.h"
int main()
{
    Widget w;
}

//输出

Impl

CAT

~Impl

对于原始指针数据成员，在 widget 对象被析构时，它的析构函数不被调用。

当 shared_ptr 数据成员时，其析构函数已被正确调用。

据我了解，在 Widget::~Widget() 中它应该生成一些 自动编码如下:

        ~Widget() {
            delete pc; // wrote by me
            
            // generated by compiler
            delete pImpl->get();
        }

为什么 shared_ptr 数据成员和原始数据成员在 widget 被销毁时有不同的行为？

我在 Linux 中使用 g++4.8.2 测试代码

================================编辑============== ================= 根据答案，原因是因为:

编译器生成的代码不是:

        ~Widget() {
            delete pc; // wrote by me
            
            // generated by compiler
            delete pImpl->get();
        }

它可能是这样的:

        ~Widget() {
            delete pc; // wrote by me
            
            // generated by compiler
            pimpl.deleter(); //deleter will be initailized while pimpl object is initialized
        }

最佳答案

因为您在头文件中转发声明“CAT”，所以您的数据类型不完整。有了这些信息，编译器就会陷入未定义的行为，并且可能不会调用析构函数。

标准是怎么说的

if the object being deleted has incomplete class type at the point of deletion and the complete class has a non-trivial destructor or a deallocation function, the behavior is undefined.

在这里你可以找到详细的解释:Why, really, deleting an incomplete type is undefined behaviour? 只需将结构声明移到类定义之前即可解决您的问题:

struct CAT
{
    std::string name;
    CAT(){std::cout<<"CAT"<<std::endl;}
    ~CAT(){std::cout<<"~CAT"<<std::endl;}
};

class Widget {
    public:
        Widget();
        ~Widget() {
            delete pc; // I know we should put this code to cpp
                       // I am just want to show the difference behavior
                       // between raw pointer and smart pointer
                       // when widget object destruct
        }
    private:
        struct Impl;
        std::shared_ptr<Impl> pImpl;  // use smart pointer
        CAT *pc;  //raw pointer
};

和输出

Impl
CAT
~CAT
~Impl

前向声明有助于加快编译时间，但在需要有关数据类型的更多信息时可能会导致问题。

但为什么它适用于智能指针呢？ 这是一个更好的解释:Deletion of pointer to incomplete type and smart pointers

基本上，shared_ptr 只需要在初始化或重置指针时声明。这意味着它在声明时不需要完整的类型。

This functionality isn't free: shared_ptr has to create and store a pointer to the deleter functor; typically this is done by storing the deleter as part of the block that stores the strong and weak reference counts or by having a pointer as part of that block that points to the deleter (since you can provide your own deleter).

关于c++ - 为什么不完整类型的智能指针数据成员和原始指针数据成员在其父级析构时具有不同的行为？，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/29062695/