The big question when implementing a runtime for a Lisp-like language in Rust is how the interpreter will interact with the GC. It looks like this project has avoided most of the complexity by not implementing actual lists, and allowing multiple references to the same value to exist only in restricted circumstances, hewing close to Rust's ownership model rather than a typical Lisp's. I wonder, is the plan to stay on this side of the space, as a lightweight language with potentially smooth FFI interoperation with other low-level/Rust code? Or to rewrite almost everything to support the high-level semantics of list-processing languages, with their emphasis on shared structure?