Re: GL slang draft

On Thursday, February 27, 2003, at 08:32  Uhr, Brent Gulanowski wrote:
>> http://www.opengl.org/developers/documentation/gl2_workgroup/
>>
> Real time work is generally more important to my projected needs, so 
> glslang looks more attractive, inasmuch as advanced shading interests 
> me at all. To speak objectively, will choosing glslang result in a 
> skewing of 3DKit toward real time applications? If so, is this going 
> to be an impediment to its usefulness?

I don't think so. The design philosophy of the 3DKit is that everything 
which is not geometry will be set via shaders. For this we need our 
shading language, probably a subset of the RenderMan Shading Language. 
These shaders will have to be mapped to the renderer, so in the case of 
OGL this might be via slang - that's why I posted this information. 
Another option would be to implement our own GL based shader, but this 
is a good approach.

<snip>

> Should not the shader language be directly driven by the choice of 
> renderer? Why should 3DKit have to choose one or the other? The 
> shaders have no impact on the scene graph, do they?

The 3DKit uses G3DShader objects which in turn use our own shading 
language to specify material properties etc. in a renderer independent 
way. So to me the question is how to implement a G3DKit Shader -> GL 
backend translator. In the case of an OGL 2.0 renderer this task could 
be greatly simplified by using glslang.

-Phil
--
Philippe C.D. Robert
http://www.nice.ch/~phip