Thank you, that's interesting. The classic Lisp machine systems are very compact by modern standards but not 2100 lines compact, although they had about two decades of continuous development to accumulate cruft (and useful features like various network protocols support and sophisticated graphics as well).
Yeah, LispM system software is orders of magnitude larger than that, but, for example, David Betz's XLISP for CP/M is 2800 lines of C, and it doesn't include any of those facilities except for an interpreter. Now, the interpreter does have some things F83 doesn't: a fully dynamic object system which supports adding new methods at runtime, named stack-allocated variables, dynamic typing, a string type, a file-pointer type with character I/O, Lisp lists, S-expression reading and printing, and a garbage collector. But it doesn't have an assembler, a memory dumper, a bytecode interpreter, a debugger of any kind, an editor of any kind, any kind of version control, any kind of multitasking, or virtual memory. For many of these, adding them in Lisp is a lot more difficult than adding them in Forth.
Probably the most influential use of XLISP is as AutoCAD's built-in scripting language, AutoLISP, but it's also the basis of Nyquist.
