单件模式的析构
第一种单件模式的实现:

C/C++ code

class CSingleton1{ private:     CSingleton1(){        cout<<"CSingleton1 constructor"<<endl;    }     CSingleton1& operator=(CSingleton1& o){}     CSingleton1& operator()(CSingleton1& o){}     public:     virtual ~CSingleton1(){        cout<<"CSingleton1 destructor"<<endl;    }     static CSingleton1& getInstance()     {         static CSingleton1  instance;         return instance;         } }; int main(){    CSingleton1 a1 = CSingleton1::getInstance();    CSingleton1 b1 = CSingleton1::getInstance();}

析构三次,可以理解,a1析构一次,a2析构一次,类静态函数里的静态变量析构一次. 但很不爽


第二种实现:

C/C++ code

class CSingleton2{ private:     CSingleton2()         {        cout<<"CSingleton2 constructor"<<endl;    }     CSingleton2 & operator=(CSingleton2 & o){}     CSingleton2 & operator()(CSingleton2 & o){}     static  CSingleton2  *instance;    public:     virtual ~CSingleton2()     {         if(NULL!=instance)                   {            cout<<"CSingleton2 destructor"<<endl;            delete instance;             instance=NULL;                   }     }     static    CSingleton2& getInstance()     {                           if(NULL==instance)                   {             instance=new CSingleton2();                   }         return *instance;     } };int main(){    CSingleton2 a2 = CSingleton2::getInstance();    CSingleton2 b2 = CSingleton2::getInstance();    delete &a2;}

这样更崩溃了,一个delete用调用析构函数,析构函数再用一个delete,然后又是析构函数......无限循环了.


哭啊,请问各位怎么用C++实现单件模式的析构!!!

[解决办法]
第二种是基于指针实现的，没法做到自动析构。所以你不应该在析构函数里再delete。
除非你再用别的类包装一下，只是那样更没意思了。
单件模式的实现未必都能做到自动析构，到底用哪种，看你自己的权衡了。