On Fri, 25 Mar 2011 11:40:47 -0700, "Ian Romanick" <[email protected]> wrote: > Previous to this patch series the mapping of GLSL built-in uniforms to Mesa > state vectors happend in the back-end. This posed a major problem for the > planned conversion of the ARB_vertex_program assembler to GLSL IR. In > assembly shaders, developers can create the moral equivalent of uniform arrays > that pull individual vector values from arbitrary pieces of state. This makes > it impossible to do name-to-slot matching in the back-end.
I like this. Among other things, this should make it possible to split the use of the builtin structure uniforms so that when someone loads, say, gl_Point.size, they don't pull in the STATE_POINT_ATTENUATION state and state flag as well.
pgpNqEng7BoEE.pgp
Description: PGP signature
_______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
