Hey, thanks for checking this out. I wanted smooth motion to keep close to the JS piece that inspired this; it seemed like it should be eminently possible to get smooth motion, after all.

And yeah, I was reminded of Bresenham's algorithm a few days back, and had a suspicion that I've basically redone it independently (and inefficiently) by doing the trig. As mentioned at the end there, the math isn't where I'm losing most time in rendering though.

The Mastodon release post (cite 0) has comments talking about doing the score recalculation only at time of collision, but the worst thing I do is use a block of RAM for the playing field and another block of RAM for the _display_ of the playing field. Copying those blocks of 20 bytes, gosh it's slow.