间接调用虚函数

运行结果是什么？

father *pf = new father();
pf->A();
pf->B();
delete pf;

father *ps = new son();
ps->A();
ps->B();
delete ps;

son *ps1 = new son();
ps1->A();
ps1->B();
delete ps1;

son *pss = new sonson();
pss->A();
pss->B();
delete pss;

运行结果

this is father::B()
this is father::B()
this is father::B()
this is father::B()
this is father::B()
this is son::B()
this is father::B()
this is sonson::B()

解释与结论

id	指针名	指针的类型	实际的类型	运行结果	解释
1	pf	father	father	this is father::B() this is father::B()
2	ps	father	son	this is father::B() this is father::B()	由virtual关键字在子类可知此处不会有多态
3	ps1	son	son	this is father::B() this is son::B()	此例中，指针的类型和对象的类型都是son，为什么最后调用的是fathher::B()?
4	pss	son	sonson	this is father::B() this is sonson::B()	由virtual关键字在子类可知此处应该有多态效果，为什么最后调用的还是fathher::B()?

解释：

在上面的测试用例中，ps1->A();和pss->A();的运行结果都是father::B()。它们都有一个共同的特点，就是指针通过A()去调用B()。

当程序通过A()调用B时，它已经处理father::A()的作用域里面了，因此此时它实际上已经不再通过ps1或者pss去访问B()，而是通过father::this。

father::this显然是father类型的指针，所以它会先去查看father::B()。它发现father::B()不是虚函数，也就不会去查vtable，而是直接调用father::B()了。这一部分的原理与virtual关键字在子类相同。

其它测试

virtual关键字在父类

virtual关键字在子类

virtual关键字在子类

完整代码

#include <iostream>
using namespace std;

class father
{
public:
  void A()  { B(); }
  void B()
  {
    cout<<"this is father::B()"<<endl;
  }
};

class son : public father
{
public:
  virtual void B()
  {
    cout<<"this is son::B()"<<endl;
  }
};

class sonson : public son
{
public:
  void B()
  {
    cout<<"this is sonson::B()"<<endl;
  }
};

int main()
{
  father *pf = new father();
  pf->A();
  pf->B();
  delete pf;
  father *ps = new son();
  ps->A();
  ps->B();
  delete ps;
  son *ps1 = new son();
  ps1->A();
  ps1->B();
  delete ps1;
  son *pss = new sonson();
  pss->A();
  pss->B();
  delete pss;
  return 0;
}