Frank A. Vorstenbosch wrote:

> IMHO, things like the basic stamp is the logical way of doing things
> -- lean interpreter, allowing for user token extensions etc.  Dunno
> whether the basic stamp uses just tokenized basic (silly) or something
> more like a stack-based system (proper).

  It«s Basic Jim, but not as we know it...

  Well, I don«t think it is really Basic at all, but it looks like Basic
before it is compiled into tokens.  It could just as easily be FORTH,
and I still have a dream of doing that ... but there is already a PIC
FORTH I«d have to study first.

  I used to think FORTH, being a multi-stack language, couldn«t be done
without a stack, but then the optimised or "assembled" versions keep at
least the top stack item in registers and can use direct or subroutine
threading instead of the classical direct.

  You think you simply *can«t* have much of a stack on a machine with a
three-cell hardware stack but once you are using token code, you are
"stacking" token pointers rather than machine instructions and for this
you can of course diddle the FSR register a lot.  Tokens or stacks is
not an "or" but an "and".

  What it really comes down to, is that there are very few ways of
performing the basic functions; looping, conditionals, subroutines on
such a sparse machine, so at the end of optimisation is is going to look
virtually the same.
-- 
  Cheers,
        Paul B.