This is a very immature, buggy, dog-slow prototype of a Lisp dialect with dynamically extensible syntax using (full) context-free grammars. Implemented in Ruby.
To fire up a REPL, enter on the command line:
> cd src
> ruby -I. cherry.rb
Then try this:
> (extend-syntax (abs Form ("|" Form "|")))
(cf abs Form ("|" Form "|"))
> |1|
Unbound variable: abs
> '|1|
(abs 1)
> (define abs (lambda (x) (if (geq x 0) x (min 0 x))))
abs
> |1|
1
> |(min 0 1)|
1
For more examples see the accompanying report in paper. There's no
guarantee that they (still) will work, though.
There aren't many built-in primitives in CherryLisp and there's not
much of a standard library. For a list of primitives, see
src/cherry/primitive.rb and src/cherry/primitives.rb. Unlike most
Lisps, CherryLisp's built-ins all use alpha-numeric names. The
hypothesis is that a nice syntax can be built on top of the basic Lisp
layer using the syntax extension mechanisms. M-expressions anyone?
The Lisp-side of the implementation is a complete rip-off of Peter Norvig's JScheme. The parser technology is based on techniques introduced by Jan Rekers in his dissertation.