如果我们有以下简单的双链表:
class Node{
//...
Node* next;
Node* previous;
int data;
};
class A{
//...
A(const A&);
~A(); //goes through list and deletes each node
A& operator=(const A&);
Node* begin;
Node* end;
int id;
};
A::A(const A& copyThis)
{
//creates a new list that represnets the same data as copyThis's list
//but is not the same list
//Example for clarification:
//if copyThis represented a list of nodes each containing the values 1-10 respectively
//this constructor would make a new list of nodes each containing the values 1-10
//and make the invoking object's members point to the beginning and end of it
//assume this does not create any memory leaks
}
A& A::operator=(const A& a) //this causes a memory leak
{
A* aCopy = new A(a); //now 'a' and 'this' should represent the same
//values but point to different lists
this->begin = aCopy->begin;
this->end = aCopy->end;
this->id = aCopy->id;
//code here that deletes list that invoking object used to point to
//assume it does correctly and without memory leak
return *this;
}
假设我们有这个功能:
void foo()
{
A a1();
//put some data into a1
A a2();
{a2=a1;}
}
我的问题是,为什么这会导致内存泄漏,因为a2
应该代表与aCopy
在operator=
内部相同的Node
列表,并且当a2
超出范围时,它会正确释放为每个节点分配的内存。
编辑:好吧,所以我在发布这个问题几分钟后才意识到,也许用new
创建aCopy
会为它所代表的Node
分配更多的内存,而且内存永远不会被释放。这是正确的吗?
编辑:我并不是在要求一个更好或正确的方式来编写这段代码,我只是要求解释为什么这会导致内存泄漏
首先,operator=需要返回T&。你正在泄露一份副本指向的内存。你根本不需要。你应该做这样的事情:
A& A::operator=(const A& a)
{
this->begin = a.begin;
//etc.
return *this;
}