Drew Northup wrote:
> Ok..., that is most certainly true... I don't think that OGL specified the
> registers or addresses though......, so it could still be a tough hack.
>
> Drew Northup, N1XIM
>
What registers are these?
If you are referring to registers on the video card - don't.
Only the driver needs to know about them (as far as I can remember) so they
don't actually need to exist.
We would need to:
fake a pci entry for the card
wrap MESA so that it looks like a video driver to windows
introduce a new target in MESA that queues messages in a buffer (in our new
'card')
convert windows GUI calls (bitblt etc.) to openGL calls
call the appropriate MESA function (applications can do this directly)
pick up the queued messages on the host and pass them on to the host openGL
implementation.
If direct buffer access is required, map it through our fake pci graphics
card. This would be OK for full screen but a b**ch for windowed access.
I notice in another thread that people are concerned about performance and
openGL only supporting X. There is (was?) an effort by the GGI project
(http://www.ggi-project.org/ ) to write an openGL implementation that would run
on any GGI supported target. This started over 18 months ago but I dropped off
the list and haven't been back so I can't vouch for the current state of it.
If I have missed anything, feel free to apply a wet haddock to the jawbone...
Bryan
