There's a REALLY cool chapter in the first edition, Chapter 12 ("Compiler Derivation") about using continuation-passing style to incrementally convert an interpreter to a (bytecode) compiler. I had never seen that approach before, and it completely blew my mind. Steele's _RABBIT_ (http://repository.readscheme.org/ftp/papers/ai-lab-pubs/AITR...), Krantz's _ORBIT_ (http://repository.readscheme.org/ftp/papers/orbit-thesis.pdf), and Appel's _Compiling with Continuations_ develop the approach further.
It was cut from later editions, for some reason. Too advanced? Too tangential to the rest of the book? Too bad.
