Window Wrapper Class, WM_NCCREATE never sent

I've created a window wrapper class, that registers a new window class, and then creates a window. However, I am unable to call my window procedure, because the static message router function never receives the WM_NCCREATE message. I always wrap my windows in this way, and have never encountered this. The only difference this time is that the class is inherited from a more basic window wrapper class. However, I have stepped through the program, and I see no reason why the WM_NCCREATE message is not being received.

If anyone has encountered something like this, give your input, and if you can be bothered going through my code, tell me and I'll post it.

The only obvious reason for not receiving WM_NCCREATE (or WM_CREATE) messages is window subclassing. In this case message handling is not initiated until after window creation so these messages, which are sent during system processing of CreateWindowEx, cannot be handled.

If you're not doing that then posting/attaching your code may help us work through the problem with you.

Nope it's an entirely new window class. I've just found that using a normal function, ie, not a member of any class, as the window procedure, works fine. Anyway here's the code.

Lol I just realized that the code is mostly based on the code from your site Ken.

At the moment the code should work, because I changed the window class to use the non-member window procedure.

Ok this is really embarrasing. Don't bother checking the code it's fine. In the static window message router, there's an if statement to check validity of the class instance pointer:

Code:

---

CAppWnd *caw=NULL;
.
.
.
if (!caw)
  return 0;
caw=(CAppWnd *)GetWindowLong(blah blah);

---

As you can see, the check comes before it's even initialised properly.

:rolleyes: :rolleyes: :rolleyes: :rolleyes:

Haha. Don't worry about it, Benny. :D

I'd like to suggest that you make sure that you call DefWindowProc for messages sent before WM_NCCREATE, such as WM_NCCALCSIZE, to ensure default system processing of these messages.

Good idea, thanks for the help.