Mon May 11 09:44:49 CEST 2009

tower of interpreters

In [1] on page 44 it is mentioned that Futamura projection practically
requires a partial evaluator that is both powerful enough to
specialize itself and simple enough to be specialized by itself.

Now, what is the relation between writing interpreters in Staapl's
macro forth to be specialized by its partial evaluator, and writing
them directly as towers of macros?  I'm confused by terminology here,
but I'm thinking practically about protocol parser specification etc..

[1] http://thyer.name/phd-thesis/