C / C++ / MFC

George_George wrote:

The design pattern is, embed a sub-object into another object (e.g. embed Goo object g into Foo object),

include <iostream>
class Goo
{
public:
  int id;
  Goo() {
    printf("Goo()\n");
  }
  virtual ~Goo() {
    printf("~Goo(%d)\n", id);
  }
};
class Foo
{
public:
  Goo g;
  Foo(const Goo& v) : g(v) {
    printf("Foo()\n");
    g.id = 2;
  }
  virtual ~Foo() {
    printf("~Foo()\n");
  }
};
void func()
{
  printf("--> func()\n");
  Goo g;
  g.id = 1;
  printf("sep\n");
  Foo f(g);
  printf("<-- func()\n");
}
void main()
{
  printf("before func()\n");
  func();
  printf("after func()\n");
}

before func()
--> func()
Goo()
sep
Foo()
<-- func()
~Foo()
~Goo(2)
~Goo(1)
after func()

George_George wrote:

What do you want to prove in this sample?

George_George wrote:

2. the order of destructors (of current instance and members) being called.
Your sample answered (1), but how about (2)?

George_George wrote:

I have looked at the result -- I think it is what you mean. What I mean is (sorry I have not made myself understood), is it Spec regulated that destructor of current object instance is invoked before destructors of member variables? Or implementation specific things?

George_George wrote:

It just mentioned in the reverse order of construction. Not mentioning the order of construction in the link you referred.

George_George wrote:

I have found the initialization order below. I think it means construction of member variables before the completion of constructor of current object. I think it also means the construction of member variables are before construction of current object.

So, the reverse order should be, destruct current object, then destruct member variables. Right?

typedef struct _MyTinyStruct {<br />
  unsigned short a;<br />
  unsigned short b;<br />
} SMyTinyStruct ;

#pragma pack( push, packSMyTinyStruct, 2  )<br />
<br />
typedef struct _MyTinyStruct {<br />
  unsigned short a;<br />
  unsigned short b;<br />
} SMyTinyStruct ;<br />
<br />
#pragma pack( pop, packSMyTinyStruct )

JBAK_CP wrote:

I've tried to handle the 'HDN_ITEMCHANGING' message...

void Cxyz::OnNMCustomdrawListFunc(NMHDR *pNMHDR, LRESULT *pResult)<br />
{<br />
	LPNMCUSTOMDRAW pNMCD = reinterpret_cast<LPNMCUSTOMDRAW>(pNMHDR);<br />
	m_list.SetColumnWidth(1,50);<br />
        m_list.SetColumnWidth(2,50);<br />
<br />
	*pResult = 0;<br />
}