Links
*****


SourceForge
===========

Psyco is hosted on SourceForge. Go to the `SourceForge project page.@http://sourceforge.net/projects/psyco`


Related projects
================

The direct sequel of Psyco is `PyPy@http://codespeak.net/pypy/`.  We are
currently working on PyPy's JIT, which uses the same basic techniques as
Psyco.  PyPy is a complete implementation of Python written in Python,
and its JIT is going to be automatically generated from the interpreter,
instead of being hand-written in C like Psyco.  This is my current
focus; I will not develop Psyco any more, beyond basic maintenance as
long as it's reasonable to do so.

If you are looking for different kinds of projects that can speed up
your programs, and if you are ready to make larger changes to your
Python code, you might be interested in the following tools:

- `Pyrex@http://www.cosc.canterbury.ac.nz/~greg/python/Pyrex/` lets you write code that mixes Python and C data types any way you want, and compiles it into a C extension for Python. This is a *compile-time* tool that uses programmer annotations in the Python source to produce a more efficient C version of your code, which you then compile with a regular C compiler.

- `Weave@http://www.scipy.org/site_content/weave/` is a tool for inline C/C++ in Python. The basic functionality lets you write C or C++ code in the middle of your Python program. It calls (and thus requires) a C/C++ compiler at run-time. On top of this you will also find functions that directly accept Python Numeric expression, or that build a regular C extension module for static compilation.

Dynamic compiler back-ends
==========================

Here are a couple of non-Pythonic dynamic compilation projects. I am often pointed to them as potentials back-ends for Psyco, to help make Psyco portable beyond the current x86 machine code. However, I consider them as not really suited for Psyco. The problem is that they all assume things about the way the code will be emitted, e.g. that we emit a whole function at a time. This model does not suit Psyco at all, where I need all the time to write a small portion of a function, run it, then compile the sequel, and so on. This said, I believe that e.g. a Dynamo-based project could indeed give very good results if applied to Python. This would just be a different project: Psyco does not really work like that. It is indirectly based on the general notion of `partial evaluation@http://www.google.be/search?q=partial+evaluation`.

- `vcode@http://www.pdos.lcs.mit.edu/~engler` is a very fast dynamic code generation system.

- `Dynamo@http://www.cs.indiana.edu/proglang/dynamo/` is a dynamic compiler and optimizer.

- `GNU lightning@http://www.gnu.org/software/lightning/` is a simple library for dynamic code generation.

- `Self@http://research.sun.com/self/` is a dynamic language for which quite some research has been done on optimization.

- `LLVM@http://llvm.org/`, the "low-level virtual machine", defines its own assembler-like format and compiles it efficiently, either normally or just-in-time.