pyLisp-NG 2.0.0

Features

• A tiny Lisp dialect, easy to integrate into other Python projects.

• Includes an interactive Lisp interpreter.

• Expression introspection specifically designed for use in genetic programming.

Introduction

pyLisp-NG is an immediate descendant of PyLisp, which had its accidental birth as a result of William Annis’ desire to build a CLIPS-like [1] syntax for a simple expert system intended for the monitoring tool, Mom. As a result of Lisp interest on the Python news group [2], William released the software and announced it to the group [3].

Several years later, the Evolver [4] and txEvolver [5] projects needed a functional programming language implementation in Python, as inspecting Python’s AST was too much of a hassle. The author wanted to distribute, process, and manage evolutionary algorithms/programs across multiple remote Twisted servers, and manipulating permutations of partial programs was much easier to integrate with Twisted if the programs themselves could be evaluated and introspected easily with Python.

pyLisp-NG inherits PyLisp’s idiosyncratic charaateristics. For example, the booleans are true and false and it doesn’t follow the convention that everything that isn’t false is true (common to many Lisps). In fact, all booleans in pyLisp-NG are based on fuzzy logic, so you can do trickier things. Macros are, quite strangely, first class objects in PyLisp which maybe novel to Lisps [6]:

lisp> (setq foo 22)
22
lisp> ((macro (x) `(setq ,x (+ ,x 1))) foo)
23

pyLisp-NG, like its predecessor, has no intent of becomming a full-fledged, robust Lisp implementation. All progress is governed solely by the interest of contributing developers; right now, that’s strictly for use as a genetic programming tool.

Installation

pyLisp-NG is setuptools-friendly; you can install it with the following:

$ easy_install pyLisp-NG

You can also get the full source code from Launchpad:

$ bzr lp:pylisp-ng

The latter option greatly encouraged for interested users, as the repository contains extra files that are not included in the ditribution tarballs (such as the test runner and test utils).

Usage

To use the interpretter from a full checkout, do this:

$ cd <branch directory> $ ./bin/pylisp-ng

If you’ve installed with easy_install, and your Python scripts directory is in your PATH, all you’ll need to do is this:

$ pylisp-ng

At this point, you’ll be able to enter Lisp expressions:

lisp> ((lambda (x) (* x x)) 2)
4
lisp>

You can use the s-expression code from Python:

>>> from pylispng import lisp
>>> l = lisp.SExpression('((lambda (x) (* x x)) 2)')
>>> str(l)
'((lambda (x) (* x x)) 2)'
>>> l.eval()
4

Building an expression one symbol at a time or by adding chunks at a time:

>>> l = lisp.SExpression()
>>> l.append('+')
>>> l.append('3')
>>> l.append('5')
>>> l.append('(* 3 3)')
>>> str(l)
'(+ 3 5 (* 3 3))'

You can examine various properties of the s-expression:

>>> l.eval()
17
>>> l.getDepth()
2
>>> len(l)
4
>>> l.getSize()
2

Known Bugs

None currently known.

TODO

• support testing .l files

• support lisp interpreter tests in doctests?

• add http://www.biostat.wisc.edu/~annis/creations/PyLisp/tests.txt to tests

• split up monolithic lisp.py file into separate files

• replace fragile interpreter that originally came with PyLisp with one based on the Python interpreter

  • graceful quit

  • command history

  • auto-completion

• add support for getting expressions by index (and thus slicing as well)

• add support for getting sub-expressions

Changes

References