Question

我正在使用std::map＆＃39; int＆＃39;作为键和自定义结构作为值。当我访问密钥时，它会创建密钥和值（如预期的那样），但是值的一次调用会被调用一次，同时它会被调用两次。我发现它真的很奇怪。

示例：

struct stStruct
{
    int some_value;
    stStruct()
    {
        some_value = 10;
        printf("\nCame in stStruct c'tor");
    }

    ~stStruct()
    {
        some_value -= 10;
        printf("\nCame in stStruct d'tor");
    }
};

int _tmain(int argc, _TCHAR* argv[])
{
    // Map of <int, struct> 
    std::map<int, stStruct> MyIntStructMap;
    MyIntStructMap[0]; // stStruct ctor called once and dtor twice
    MyIntStructMap[0]; // neither stStruct ctor nor dtor called
    printf("\nValue of some_value in stStruct %d", MyIntStructMap[0].some_value);
    // Expected -10 above as dtor was called twice
}

输出是：

Came in stStruct c'tor
Came in stStruct d'tor
Came in stStruct d'tor
Value of some_value in stStruct 10

这真令人沮丧。特别是如果在结构中有指针并分别在它的ctor和dtor中分配deallocate内存，则代码崩溃（因为同一指针的删除发生两次）。

此外，尽管代码some_value -= 10;被调用了两次，我仍然不明白，上面示例中some_value的值如何仍然是10？

我在Windows上使用VS2010。

Answer 1

这是因为在使用l值索引运算符进行地图访问期间正在执行编译器生成的c-ctor。

检查此代码

struct stStruct
{
   int some_value;
   stStruct()
   {
       some_value = 10;
       printf("\nCame in stStruct c'tor");
   }

   stStruct(const stStruct& oOrg)
   {
       some_value = oOrg.some_value;
       printf("\nCame in stStruct copy c'tor");
   }

   ~stStruct()
   {
       some_value -= 10;
       printf("\nCame in stStruct d'tor");
   }
};

int _tmain(int argc, _TCHAR* argv[])
{
   // Map of <int, struct> 
   std::map<int, stStruct> MyIntStructMap;
   MyIntStructMap[0]; // stStruct c'tor will be called once and d'tor will be called   twice
   MyIntStructMap[0]; // stStruct c'tor or d'tor won't be called
   printf("\nValue of some_value in stStruct %d", MyIntStructMap[0].some_value); // As d'tor was called twice, ideall it should print value -10

   return 0;
}

产生

Came in stStruct c'tor
Came in stStruct copy c'tor
Came in stStruct copy c'tor
Came in stStruct d'tor
Came in stStruct d'tor
Value of some_value in stStruct 10

在以结构为值的STL映射中，结构的析构函数被调用两次

1 个答案: