I have readed this file:
http://oss.sgi.com/projects/ogl-sample/registry/doc/GLSLExtensionRules.txt
And then I remembe that in Cg there is a type half for half float.
Then I have a look on GL_NV_half_float specification to see if there half float are available on GLSL with this extension.
It wasn’t the case, but I still have question about how this half type could be add into GLSL according GLSL extension rules.
The type itself should be named __halfNV and we could add also something like __hvec2NV, __hvec3NV and __hvec4NV. It could be useful to apply a noramlization on these vectors.
Is it need to create a function like normalizeNV or the build in function normalize could be overloaded by GLSL extension?
Also is nVidia card support half float, why they didn’t update the GL_NV_half_float specification?