Finally after a couple more mapping sessions I feel like I've turned a corner with the map. The end of the excruciating rock/terrain triangle brushes seem to be in sight, only a little further to go. It was a big relief to get enough done that I could seal the map up and compile it again for the first time in over a month. Good to see that the compiler still isn't a problem. Up to 44k+ clipnodes now, but that's totally unoptimised. Now it's becoming apparent that the long vis times required (~48 hours I'm guessing, still waiting...) will be a bit of a problem for testing. I'll have to hope that a fast vis will be good enough for the testers, otherwise there could be a bit of a problem making the QExpo deadline. Fingers crossed.

I've uploaded another update to TyrQuake (v0.54) as well. This includes the removal of the gl_texsort 0 rendering path, which isn't really an advantage on any cards I know about (and usually a big disadvantage on slower cards). Like I said with my May update, this also makes sure the particle rendering looks correct. Other than that, nothing else really user-visible; just more juggling of the code internals as usual.