Thanks for the livecoding link. I assume there's some overhead in constantly checking for and applying code changes - that's probably the reason for the lower frame rate.
Good point, I hadn't thought about the fact they might be polling for changes. Might be better if livecoding used an event based model to check for updates...
Or if it works at all like Scrubby (http://nornagon.github.com/scrubby/) which was on HN a couple days ago, it might rewrite all constants into global lookup table indexes, which surely wreaks havoc on performance.
How do you suppose event-based models are notified of changes? Ultimately you are still polling, you're just abstracting it out of your higher-level design.
