>>27
OpenGL 4.2 or 5.0 or whatever they decide to call it, that is, the next upcoming revision to OpenGL, will contain a full-blooded modern replacement for display lists that will support parallelism: you will be able to construct draw call lists across multiple threads. Direct3D11 has supported this since 2009 with multi-threaded deferred render contexts.
I agree that Display Lists are bloated overkill, but why did they deprecate them before introducing a replacement for static geometry?
Also, the FF-P was shit, get over it. It had a lot of call-overhead and had shitty fine-grained state management which limited overall throughput. It's fucking trivial to write just as optimized emulation shaders. If you can't do that, you suck. Even John Carmack, a long-time OpenGL advocate, came out last year and declared that Direct3D had been better than OpenGL since Direct3D9, for largely these reasons.
If a new pipeline stage gets integrated in the graphics hardware, programs using a future-proof FF-P can take advantage of it without needing an update. We've seen that before with hardware T&L. Of course, OpenGL's FF-P isn't future-proof anymore.
Also, I compared the performance of FF-P vs. shaders (GLSL) with orthogonal textured quads on an old R300 with the free radeon driver, and the FF-P gave me a few more FPS. Perhaps the Gallium3D shader compiler sucks and I gotta try assembly.