Whenever my D3D window is completely covered, the mouse movement is screwed up. Is there a windows message or function that I can use to determine if a window is completely covered? Someone please help! I''ve been trying to find an answer for about a week. Please provide code if you can. Thanks.

This should not be a problem. In fact, a completely covered window should not affect mouse movement at all.

What sort of mouse-oriented code do you have in place? In particular, are you attempting to capture/release the cursor? Hide/show the cursor? Are you using DirectInput?

[edited by - darrell l on November 6, 2002 11:33:49 PM]

lucid, you probably have your code inside the main
window loop. What happens when your dx window is
completely covered is something like your app
steals almost all the computer''s resources.
Everything get''s garbled, not just mouse.
I got the same thing !

I do not quit understand yet myself exactly why it
is so, but here is a simple way to handle the problem:

Just make the program wait a few milliseconds after
the completion of a frame, f.inst right after the
device->present().

I do this by applying a simple function of my own:

void delay(int ms){
HANDLE h=CreateEvent(NULL,true,false,NULL);
WaitForSingleObject(h,ms);
CloseHandle(h);
}

Then my "gameloop" is somewhat like:

while ( ''peekmessagestuff'' )
. render a lot of stuff
. dx8device->present(0,0,0,0)
. delay(5)

Now you can cover or hide your multimillion dollar
game without problems :-)

I think there must be an even better way to handle
this though. This way at least is very simple ...

My program consists of a spinning triangle. I want the triangle to spin even when the window loses focus. I am not doing any fancy mouse stuff and I am not using DirectInput. I cannot use the WM_ACTIVATE message, since that will stop the spinning as soon as the program loses focus.

The program works fine when the window is in the background or even when it is 99% covered with other windows. As soon as the window is 100% covered, I get the stalling problem that alfmga referred to. Someone else gave me a solution similar to Bratsj''s solution using a simple call to Sleep(milliseconds); While this solution works, it''s not very elegant and probably not the best way to do it.

Basically, my program thread needs to go to sleep instead of hogging the CPU. The difference between PeekMessage and GetMessage is that PeekMessage returns NULL if there are no messages in the queue while GetMessage puts the program to sleep until a message is received.

So code like this would solve my problem:

if(windowTotallyHidden)
GetMessage(...);
else
PeekMessage(...);

The problem is determining when the window is hidden... :o(

As a side note, I am bothered by that fact that I have to do something special to get proper behavior. And I am surprised that every tutorial I have downloaded has this problem. As far as I''m concerned this is a bug with DirectX or Windows... OpenGL does not have this problem.