🚀 go-pugleaf

Hi Gary,
You're right, my error was that I didn't actually write text in the WM_PAINT
message
handler... But what intrigued me is that text was displayed all the same...
In fact, it's the
SetWindowText function that drew the text. I guess it uses a function like
TextOut cause
it doesn't call WM_PAINT to display the text. So, I tried to intercept the
WM_SETTEXT
message in order to prevent Windows from displaying the text out of the
WM_PAINT
handler. Here is the new labelProc :

LRESULT CALLBACK labelProc(HWND hwnd, UINT uMsg, WPARAM wParam, LPARAM
lParam)
{
 switch (uMsg)
 {
  case WM_SETTEXT:
   SetWindowLong(hwnd, GWL_USERDATA, (LONG) strdup((char *) lParam));
   return 0;

  case WM_PAINT:
   HDC hdc = BeginPaint(hwnd, &ps);
...
   char *text = (char *) GetWindowLong(hwnd, GWL_USERDATA);
   DrawText(hdc, text, -1, &rect, DT_LEFT);
...
   EndPaint(hwnd, &ps);
   return 0;
 }

 return CallWindowProc(oldLabelProc, hwnd, uMsg, wParam, lParam);
}

The SetWindowLong function is the only method I found (with SetProp) to
store the
text of a label and get it again later. Since the label procedure is unique
for all the
labels, I couldn't store the text in a static variable : I needed a
different place for
all the labels. Though, I don't know when to free the string duplicated
(with strdup).

About the multiple labels, of course it prevents me from using the
CTLCOLORSTATIC
methos because it forces me to put lots of "case" statements in the switch
statement...
Since all variables tested in the "case" statements will be HWND, I don't
see any
way to determine its associated color in one instruction... I mean, I'm not
certain
that all these HWND's will form a serie (like 41001, 41002, 41003, etc).

Have a nice day,
Mark