I did once write a 2k-line bytecode interpreter and runtime for R4RS Scheme -- the compiler, library, and debugger were another 2-3kloc of Scheme. I even used it as my main hacking platform for years. But I didn't post it here because:
* It depends on the Boehm garbage collector.
* (This is embarrassing.) It stopped working after some version bump of GCC, and the cause is not at all obvious to me. Since I don't Scheme much anymore I gave up.
No blame, but I'm curious "why you don't Scheme much anymore". Was it due to Python, Common Lisp, etc?
I switched from (Chicken) Scheme to Common Lisp (and, in parallel, from Python to Lua), and, having read a lot of old code of yours, I'm curious about your choices - you seem consistently sensible and pragmatic. (Sorry to put you on the spot.)
Thanks! Partly Python's gotten more tolerable as a language, partly I'm doing more things needing libraries, partly I mostly code inside https://github.com/darius/halp these days and it doesn't have a Scheme mode so far. I do have a couple of recent Scheme projects up on github though -- optilamb and selfcentered.
Very impressed with LuaJIT2, btw -- I'd like to do more with it.
Would compiling Lisp code count against the challenge's LOC limit?