I did most of the fine-tuning that I mentioned in the last email.
Lighting now works. I replaced the view frustum culling algorithm with
a more efficient version. Course edges should now look ok.
The terrain is now rendered using compiled vertex arrays; since this is
how Quake 3 does things I'm hoping that it will speed things up on
certain cards/drivers (it makes no noticable difference on my Voodoo3).
HOWEVER, when I do "tuxracer -indirect" I get a core dump on
glLockArraysEXT(). Can somebody with a card that does indirect
rendering (e.g. TNT/TNT2/nVidia) please try this and let me know if they
get the same thing?