Not a lot to report this week on the development front. I’m still tossing ideas at the wall and looking at what sticks, and writing long essays to myself about why they might be sticking. Slowly, a picture emerges.

Testing proofs of concept in a hacking project has worked much better than trying to code the actual game project from the start. I feel more productive when I’m working on production code, but it’s a false economy, as I wind up rewriting most of the code and introducing plenty of bugs along the way. Better to do the rewriting in notes and hack modules that I can toss out.

I mentioned HTML5 audio a couple posts back, and specifically how I didn’t think the callback model was going to be performant when combined with WebGL. In retrospect, it might have been a good idea if I’d actually tested that assertion first. I’m happy to report that I’ve tested it for my own use cases, and I was wrong. I don’t notice any dropoff in framerate from generating raw audio on the fly, nor do I experience audio dropouts from using WebGL.

I still like Webkit’s patch-and-plug model, of course, but I no longer believe it to be the only usable one.