Ethan

I implemented the first optimization I wrote about yesterday. Oddly enough, it loses time (1.2 seconds over a 24 second run)! I think what's happening is that the empty space that isn't ordinarily rendered into is so big that when it gets merged into a big sprite, it goes slower. I guess I'll re-benchmark when Carl's finished populating the level backgrounds, but there are other levels with sufficient empty space that this might be a losing optimization. That sucks, because that will also invalidate the other optimization I had planned, which means I don't have any clever ideas on how to save big. I'll have to start studying the code and maybe wrest some small gains from my knowledge of Python, but I'm uneasy.

I'm leaving the merge-bg code in for now, but disabling it.

Ethan

P.S. I tried filling in all the empty spaces. With that change, the optimization is good -- 25 fps versus about 20 -- but it isn't enough anyhow.