The patch actually fixes every app which dares to use ARB_draw_instanced. The only apps the very presence of the extension doesn't break are piglit tests. It's clear from bug reports that not only is the current implementation of the extension absolutely useless in reality, it also prevents users from using certain apps (isn't it great?). The standard is worthless if it doesn't let you be compatible with the rest of the world and, it this case, it sort of does by this line:
EXT_gpu_shader4 or NV_vertex_program4 or OpenGL 3.0 is required. Well there certainly is a reason it depends on EXT_gpu_shader4: gl_InstanceID. And it nicely matches the reality. I don't insist on committing the patch, I can disable the extension from a driver. However I thought other drivers would need the same fix. Marek On Sat, Apr 16, 2011 at 4:18 PM, Brian Paul <brian.e.p...@gmail.com> wrote: > On Sat, Apr 16, 2011 at 5:06 AM, Marek Olšák <mar...@gmail.com> wrote: > > From GL_ARB_draw_instanced: > > > > EXT_gpu_shader4 or NV_vertex_program4 or OpenGL 3.0 is required. > > > > which means we can't advertise it. :( Well at least apps won't be looking > > for gl_InstanceID. > > I believe that the GL_ARB_draw_instanced spec is wrong about that. > There's really no reason why instancing depends on those extensions. I > think that Ian has reported this issue along with gl_InstanceID vs. > gl_InstanceIDARB to the ARB so the spec may be changed soon. > > Typically, ARB extensions do not directly depend on EXT extensions. > > -Brian >
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
