GSOC (proposal)

Stefan Dösinger stefan at
Mon Feb 25 09:11:08 CST 2008

Am Montag, 25. Februar 2008 11:32:03 schrieb David Adam:
> Here are my wishes for GSOC:
> D3DX dlls begin to be implemented. It seems that this could be a good GSOC
> to implement more feature of this dll, in particular the shader compiler.
> In the view of wined3d, I think about vertex blending, since there are tons
> of game using this feature.
> Stefan, is this doable in the scope of GSOC?
I want to write a shader compiler myself withhin the next months as an 
university project(otherwise I have to do some java junk). I should know more 
about this within a few days.

Implementing Vertex Blending via vertex shaders means writing a replacement 
shader for the fixed function vertex pipeline. Henri has started such a 
thing, but never completed it due to the lack of time and need(as far as I 
can see "tons of games" means one game and 4 directx sdk samples). Writing 
the replacement shader shouldn't be that much of a problem. The algorithms 
are well known and can be read in MSDN, the opengl books or pretty much any 
book on 3D rendering math. The bigger problem is integrating this work into 
wined3d in a proper way. I don't know if that is doable for a soc student, my 
last estimation on that topic wasn't that good. Henri, what do you think?

What I want to make clear: To implement Vertex Blending we have to replace the 
*whole* vertex pipeline. We can't just implement vertex blending. We also 
have to deal with texcoord generation, regular transformation and lighting, 
etc, etc, etc.

> Sound should be improved also, since there are games that still crash, in
> despite of M. Lankhorst's great job ( dmime, quartz,.... dlls problem).
Those are Internet Explorer / Windows Media Player DLLs I think, and not 
directly related to winmm/dsound.

> And the last but not the least :D: fix the infamous warping mouse bug. I
> know that it is mainly a problem with X server, but maybe a wine develooper
> could propose  a GSOC to Xorg guys to fix this.
We certainly have to talk to the Xorg guys about that. A few things I heard 
was that Vitaliy says opening the core pointer with Xinput is reasonable, and 
that he once had a one-liner hack in the X server to make that work. Going by 
that it would be very simple work, but who knows what fun shows up in the X 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url : 

More information about the wine-devel mailing list