Sat Nov 20 08:21:12 EST 2010

The Forth -> module parsing is too complicated

Reason: the `expand' re-structuring after a require statement.

Maybe it's simpler to expand from Forth (concrete) syntax straight to
Scheme module syntax instead of going through forth-begin.

Or: is the split between lexing and parsing really necessary?  It even
complicates standardizing Forth.

I think it's important to keep in mind that the basic form should be
an s-expression that integrates well with the rest of Racket.

What about this:

   * Write a non-extensible flat Forth lexer that compiles to (module
     ...) form.

   * Is it possible to do this in a way that is extensible?  I.e. can
     we have a read-time Forth running?

   * Can the same code be used on-target?

So, the point really seems to be to write a stand-alone Forth and then
stub it out to plug in the compiler/macro part.

Can we start out with something like eForth and move on from there?