Q  
When I use a modeless dialog in the SDK, I have to insert a statement in 
the main loop like this: 

while(GetMessage(&msg,NULL,0,0) { 
    if(!IsDialogMessage(hdlg,&msg){ 
        TranslateMessage(&msg); 
        DispatchMessage(&msg); 
     } 
} 

 
      My question is: why do I need this IsDialogMessage call? Isn't a 
modeless dialog just another window like the main window? So without 
this addition can't the message be dispatched to the dialog? If I have 
to add this statement, does that mean that if I have several modeless 
dialogs I have to add a separate IsDialogMessage call for each one? If 
so, how does MFC handle this? I can't touch the main loop in MFC, so how 
do I add IsDialogMessage to it? 
Hua YuanA  
Yikes—that's more than one question! But let me try to answer. First of 
all, part of the confusion is the name. The purpose of IsDialogMessage 
isn't to check whether the message is for the dialog; it's to translate 
the message. A better name would be TranslateDialogMessage.
      As with TranslateAccelerator in the previous question, IsDialogMessage 
translates messages for modeless dialogs. For example, it maps 
Tab and Shift+Tab to the next/previous control, and maps the arrow 
keys to their respective operations in a dialog. You can think of it as 
translating the message based on a special built-in accelerator table just 
for dialogs. In fact, you don't have to use IsDialogMessage for dialogs—you 
can use it for any window that contains controls to get the same keyboard 
behavior as a dialog. But, since the first argument of IsDialogMessage is the 
HWND of the dialog, you do in fact have to call it for every modeless dialog 
your app has running at a given time. Normally you'd have a global HWND 
variable, g_hDlg, that you set when you create the dialog. 
      That explains why you need IsDialogMessage, and answers the question, 
"Do I have to add one call for each dialog?" The next question is: how does 
MFC handle this? As I described for the previous question, MFC uses a 
sophisticated routing scheme using the virtual function CWnd::PreTranslateMessage, 
which lets any window hook itself into the main loop without modifying the 
loop itself. The default implementation for CDialog::PreTranslateMessage 
looks like this: 

BOOL CDialog::PreTranslateMessage(MSG* pMsg) 
{ 
    •••
// many lines of intimidating code 
    •••
    return PreTranslateInput(pMsg); 
} 

 
In other words, CDialog passes the buck to yet another virtual function, 
CWnd::PreTranslateInput: 

BOOL CWnd::PreTranslateInput(LPMSG lpMsg) 
{ 
    if (/* keyboard or mouse message */) 
{ 
        return IsDialogMessage(lpMsg); 
    } 
    return FALSE; 
} 

 
      As you might guess, CWnd::IsDialogMessage is a wrapper that calls 
::IsDialogMessage with m_hWnd as the HWND. The result of all this is that 
in MFC, each dialog translates its own input. If you have five dialogs 
running, each dialog's PreTranslateMessage calls IsDialogMessage—so you don't 
have to. Pretty nifty! In MFC, you don't ever have to worry about 
IsDialogMessage—it gets taken care of automatically.

• Modeless dialog boxs - Бяша 02.05.02 03:00 [620]
  • решение - ggg 02.05.02 12:15 [933]
    • действительно, thx! (-)  - Бяша 02.05.02 16:54 [926]
  • from MSDN - + 02.05.02 03:56 [1523]
    • то есть другого пути нет. жаль. (-)  - Бяша 02.05.02 05:02 [960]