Fri Apr 24 20:09:03 CEST 2009
maybe it's time to rewrite the whole thing... one of the problems i
had is to have to work around limitations of the parser..
maybe it should be made to parse forth in the first place, then limit
it to be able to do simpler things too?
the basic problem is a parse from linear code -> dictionary. this is
solved using prefix parsing keywords.
so, what is needed is a representation of a dictionary, somewhat like
the chunk-collecting compiler in comp/compiler.ss
another thing to mention is (delimited) continuations. upto now this
was problematic because of the way code was mapped to lambda
expressions. it might be better to work directly in cps, since that's
a lot closer to the way forth works.
the code fragment (a b c)
start in rpn/new.ss