I want to revisit my LISP interpreter someday because I have a few ideas about how to implement it more simply. The weird thing, which I only realized after the fact, is that it's less like "building a LISP" and more like "building a LISP machine". One thing I'd like to try for that system is a shared "memory bus" that can be used to allocate and fetch data - right now each node in the system has data-fetching logic built into itself, and it's massively complicated as a result.
