Update: It might not be ready by the end of August. I just returned from a week-long trip, and I'm now busy rewriting the whole Fexl interpreter in terms of pure combinatorics (see http://fexl.com/#combinatorics ).
The interpreter understands only the two fundamental combinators S and C. Although it is possible to define high-order combinators in the interpreter using function pointers, I am limiting myself to S and C to minimize the code size and illustrate the principle that all computable functions can be defined as applications of only S and C.
So, for example, even the Y combinator itself is defined in terms of S and C as:
\I = (S C C)
\Q = (S (S (C S) C) (C (S I I)))
\Y = (S Q Q)
I'm now doing some bootstrapping, eliminating code formerly written in Perl. I'm writing the Fexl parser in Fexl, and manually converting it to S and C. I can then eliminate the parser written in Perl.
The interpreter understands only the two fundamental combinators S and C. Although it is possible to define high-order combinators in the interpreter using function pointers, I am limiting myself to S and C to minimize the code size and illustrate the principle that all computable functions can be defined as applications of only S and C.
So, for example, even the Y combinator itself is defined in terms of S and C as:
I'm now doing some bootstrapping, eliminating code formerly written in Perl. I'm writing the Fexl parser in Fexl, and manually converting it to S and C. I can then eliminate the parser written in Perl.