WINCORE.CPP Assert Message

Derek F. Groft - UTL -- groftde@ebs.ac.com
Friday, September 27, 1996



Environment:  VC++ 4.2, NT 3.51, NT 4.0

The code that I use to subclass the Edit box portion of a Combo Box has
been working fine on NT 3.51 for some time now.  The code below, taken
from my PreSubclassWindow() handler,  illustrates what I am doing.

	m_pEdit = new CMyEdit;

	POINT p = {1,1};
	HWND  h = ChildWindowFromPoint(p)->GetSafeHwnd();

	m_pEdit->SubclassWindow( h );

We now have some users on NT 4.0 and the last line causes an ASSERT
on line 308 of WINCORE.CPP.  The MFC code is as follows:
	
BOOL CWnd::Attach( HWND hWndNew )
{
	ASSERT( m_hWnd == NULL );  // only attach once.
	ASSERT( FromHandlePermanent( hWndNew ) == NULL );  
	    // must not already be in permanent map

	if ( hWndNew == NULL )
		return FALSE;

	etc.
}

The second line of this function is line 308.  

Does anyone know what changed in NT4.0 to cause this to happen or what
I should do about it??

---

Mike Blaszczak -- mikeblas@nwlink.com
Sunday, September 29, 1996

At 18:30 9/27/96 -0500, groftde@ebs.ac.com (Derek F. Groft - UTL) wrote:

>Environment:  VC++ 4.2, NT 3.51, NT 4.0

>The code that I use to subclass the Edit box portion of a Combo Box has
>been working fine on NT 3.51 for some time now.  The code below, taken
>from my PreSubclassWindow() handler,  illustrates what I am doing.
>
>	m_pEdit = new CMyEdit;
>
>	POINT p = {1,1};
>	HWND  h = ChildWindowFromPoint(p)->GetSafeHwnd();
>
>	m_pEdit->SubclassWindow( h );

This isn't very good code. (Unfortunately, you probably got the idea from a
sample shipped or article provided by MS.)

In Windows NT 3.51, you were using the old shell.  In Windows NT 4.0, you're
using the new shell.  The new shell draws combo boxes that look different than
they do in old shell.  But, unfortunately, this code is  tightly bound to the
assumption that combo boxes are drawn a particular way--that there's some edit
control at point (1,1) relative to the origin of the combo box.

In Windows NT 3.51, that assumption was true-- in Windows NT 4.0, that
assumption isn't true.

>We now have some users on NT 4.0 and the last line causes an ASSERT
>on line 308 of WINCORE.CPP.  The MFC code is as follows:

This is why I think it's so much more important to understand _how_ and _why_
code works than to just paste examples around.

This assert means exactly what the comments say: you can only attach a
HWND to one permanent CWnd object.  Your call to SubclassWindow() asks
MFC to attach a CWnd object to a window handle--that is, the implementation
of SubclassWindow() calls CWnd::Attach().

You're not getting the window you thought you were getting.  The assumption
that the edit control is at that certain physical position breaks in new
shell version of Windows. It turns out the window that's really _is_
at the physical place you've specified has already been attached and
that means you get the assertion.

>Does anyone know what changed in NT4.0 to cause this to happen or what
>I should do about it??

You should find a better way to find the edit control window.  You should,
I think, find a way that isn't based on the physical appearance of the
Window and, instead, depends on the logical relationship of the windows
involved in the control.

You could have figured this out yourself by examining the window handle
returned from your call to ChildWindowFromPoint() call. You're not
getting a handle to the window you thought you were getting.
.B ekiM
http://www.nwlink.com/~mikeblas/
Don't look at my hands: look at my _shoulders_!
These words are my own. I do not speak on behalf of Microsoft.

---

| Вернуться в корень Архива |