lab 26 - greenspun's rule
NAME
lab 26 - greenspun's rule
DESCRIPTION
Let me state up front, I do not know much about lisp. I've always been aware of it but never saw it's benefits so never used it. However, recently I've begun to read more about it, particularly the essays by Paul Graham, and his book On Lisp (which I'm still reading, on and off). I went a bit further and read Steele and Sussman's paper The Art of the Interpreter. Also, as a symptom of my Mathematica envy I became curious about symbolic programming languages, computer algebra systems, and how these are based on lisp dialects. I started playing with Maxima and yacas and also Mockmma
And so I think I finally got what lisp was about. (I still have much more reading to do. It's funny how a large and rich subculture within computer science suddenly opens up to you when you read about its core language. I have to ask myself, How could I not have noticed this for so long? The same surprise occurred to me when reading about smalltalk. It all makes me feel quite stupid and parochial.)
As an exercise in understanding lisp better I turned Inferno shell into a lisp dialect. I wrote a builtin module that provides an eval command and the seven primitive operators: quote, atom, car, cdr, eq, cons and cond.
I followed approximately the code in Paul Graham's paper The Roots of Lisp, and the interpreters described in the Steele and Sussman paper.
% load ./lisp0.dis % eval {atom `a} t % eval {atom `{a b c}} nil % eval {car `{a b c}} a
Functions can be defined by assigning to a variable in shell. Scope within eval is dynamic.
% subst={{x y z} {cond {{atom z} {cond {{eq z y} x} {`t z}}} {`t {cons {subst x y {car z}} {subst x y {cdr z}}}}}} % eval {subst `m `b `{a b { a b c} d} } {a m {a m c} d}
The backquote is used as an abbreviation for the quote primitive.
In the builtin I hijack the parse tree and read or transform it for each primitive operator. I also use the shell Context as my symbol table for function parameters, definitions and variable values.
CONCLUSION
The attached file is the first of a few versions I'll provide as I add more features as described in Steele and Sussman, in particular lexical binding.
I wonder whether there is more value to having a dialect of lisp within the shell or whether a proper lisp implementation like scheme should be linked into the inferno VM or written in limbo. The shell already has a good interface to the whole system and mixing programming paradigms may have advantages. The problem is we may realize Greenspun's rule by being ad hoc, informally specified, bug ridden and slow.
Comments