computation, game design, and experimentation

front page | about | archives | code dump | c.s. for mere mortals | tags | rss feed

A proposition for a small purely functional, typeless language

October 5, 2012
tags:  axiofunctionallanguageTuringlambda

So, one of my favorite pasttimes is language design, and a few weeks ago I whipped up the following grammar for a purely functional, typeless language, which consistes solely of lambdas and symbols. It's essentially the \(\lambda\)-calculus, with global, immutable binds (much like substituting a symbol).

Anyways, here's the grammar:

lam ::= (\ var1 ... varn :: exp)

exp ::= lam
      | var
      | (prim exp1 ... expn)
      | (exp1 ... expn)

prim ::= #bind | #halt

Aside from the grammar, comments are defined with double slashes and run to the end of the line, C++/Java style.

The two primitives have the following meaning:

(#bind sym val) => binds sym to val in the global symbol table
(#halt exp)     => evaluates exp and returns that value to the 
                   calling context (OS level) as a string

As I mentioned, it's small. Masochistically small, it appears. There's also absolutely no concept of I/O aside from return values. But it fully supports the \(\lambda\)-calculus, and therefore is Turing complete.

However, in implementation, a few things would be pre-bound to make the programmer's life simpler. These would not be part of the grammar, and therefore not a part of the syntax of the language, as they would be stored like any other bound symbol internally. There would be one exception to these pre-bound values, and that's the storage of numbers. Church numerals would be auto-converted on the fly, and then cached into the global bind table.

Semantically speaking, functions would be evaluated identically to the \(\lambda\)-calculus. Also, the interpreter would perform automated currying internally.

(\ a b :: b a) => (\ a :: (\ b :: b a))
(\ a b :: b a) someval => (\ a :: (\ b :: b a)) someval => (\ b :: b someval)

Here's an example program in this small language, which I'd like to call axio:

// fibonacci generator
// supporting functions
(#bind succ (\ w y x :: y (w y x)))
(#bind + (\ a b :: a succ b))
(#bind phi (\ a b f :: f a b))
(#bind sigma (\ x :: phi (x false) (+ (x true) (x false)))) 
(#bind fib (\ n :: (n sigma (phi 0 1)) false))

// fib
(fib (fib 5))

// exit

This would calculate the 5th Fibonacci number, let's call it f_5, and then would calculate the f_5th Fibonacci number.

There really isn't much purpose to this language, especially given the complete lack of I/O, other than it may be a neat challenge to implement, and an excuse to write a state-machine interpreter.

Ah well, that's axio.

Thanks for reading!

blog comments powered by Disqus