hamelin/catty

Concatenative programming integrated in Python

catty -- Concatenative programming integrated in Python

Concatenative programming is a method and coding style that emphasizes functional composition through implicit parameter passing and postfix notation. It is often understood as stack programming, mimicking the constraints of stack machines. While it carries some hipster cred, it is popular in development circles involving dedicated hardware. Its most popular incarnation is the FORTH language; the Factor language and platform provides a more modern programming system over its principles.

This project is an attempt at integrating a concatenative style and approach to the Python language and software ecosystem. Its present incarnation is somewhat clumsy as it parasites the Python syntax: a catty program is valid Python code. The catty semantics also attempt to leverage the Python built-ins and library as much as possible. Here is a taste:

from operator import add, mul
from catty import reduce
from catty.words import dup


result = reduce([5, 6, dup, mul, add])
print(result)  # Yields [41]

So this early take yields the answer to life, the universe and everything, but it's off by one. Should be fixed in next version.

The interpretation is performed in a Python code loop, so speed cannot be expected. However, the interpreter in its current form optimizes tail calls, enabling arbitrary recursive applications:

from operator import mul
from catty import reduce
from catty.words import dup, tuck, fork, apply, Word

factorial_ = Word(  # prod, n -- prod, n-1
    dup, 0, eq,
    [],
    [dup, tuck, mul, swap, 1, sub, apply.factorial_],
    fork
)
factorial = Word(  # n -- n!
    1, swap, factorial_
)
print(reduce([5, factorial]))  # Yields [120]

This project is meant for incremental exploration of interactions between concatenative and applicative programming styles. It will not shy away from changing completely as it progresses. Current plans involve digging further into tapping into namespace control through function definitions, replacing the interpretation loop with Python bytecode generation, and further strengthening integration so that concatenative and regular Python code can live side by side.