Hi Drew,
To finish off this thread, I'd say we're now having a semantic disagreement.
True,
DirectX won't run without the kernel dll and uses services in it. It can be
viewed
as running "on top". I look at it a little differently. There is nothing
between the
graphics hardware and the DirectX driver, so from this viewpoint DirectX
isn't "on
top" of anything. Just a different perspective on the same picture.
On the DirectX/GDI issue, in Win2K DirectX/3D is implemented in the display
driver dll,
so the two can share functionality. They don't have to and there is no
"hook", but it
only makes sense to use a common function where possible. In NT (pre-2K)
only
DirectDraw is supported, which is hardly worth the bother. In Win9x, the
display
driver has a "hook", but this is only to tell Windows that DirectX is
supported and
the name of the dll that supports it. DirectX drivers in Win9x will share a
lot less
functionality than in 2k.
Hope this helps somehow...
Regards, Colin
> OK.., well thanks about the info on the ring0 code especially...... As for
> the getting around laterally & not being dependent on the Win32API, my
> conversations with the Win2k debug manager would point in other directions
> (granted, he had to ask what VPN meant..., but when you spend that much
time
> in front of a terminal it is nice just to get to talk to real people!!).
> These being 1) DirectX will not run without the kernel*.dll/rundll*.*
> services in place (be nice if it ran on DOS alone though...), 2) in the
most
> pure sense, Win32--since it isn't really an operating system, has to deal
> with the problems of command.com; however it is the existence of some of
> these holes that allow things like DirectX to talk to the hardware without
> using kernel*.dll/rundll*.* as an intermediary. And yes, I just did a
> find.., there is no longer a kernel32.exe. And as for not using the GDI,
it
> was my understanding that DirectX has a hook into the GDI code that allows
> it to take over certian functions when the application (also hooked to the
> GDI) supports that particular implimentation (thus being a hell of a lot
> faster).
