================================
PyPy's Python Interpreter status 
================================

What this talk is about
=======================

* Why we work on PyPy?

* What you can run on top of PyPy

* How fast is PyPy?

* Details about coming 1.1 release

* Questions and Answers 


PyPy - user motivation
=======================

* use Python rather than C for performance 

* have a more speedy, resource efficient interpreter 

* support more programming paradigms 


PyPy - motivation
=================================

* CPython is nice, but not flexible enough

* IronPython, Jython - bound to the specific VM

* Psyco and Stackless Python hard to maintain

PyPy: generating Python Interpreter 
=====================================

* **high level Python specification**! 

* layer GCs, JIT, Stackless atop the spec 

* **generate interpreters** for targets 

.. image:: pypy-multitarget.png
   :scale: 50
   :align: center

Brief history of PyPy
==========================

* first sprint 2003, about 30 more by now

* CPython/Psyco/Jython/Stackless developers participating

* MIT-License, more sprints

* EU Research project 2004-2007

* 2007-now - open source project

* some google sponsoring

Getting Production ready
==========================

* we worked a lot on running
  existing applications on top of PyPy

* sometimes requiring to change applications slightly

* especially refcounting details tend to be a problem

::

  open('xxx', 'w').write('stuff')

CTypes
======

* official way to have bindings to 
  external (C) libraries for PyPy

* can handle i.e. pysqlite-ctypes, pyglet, pymunk or Sole Scion,
  almost whatever....

* contribution to original ctypes
  (better errno handling, bugfixes, tests...)

* part of google sponsoring

* note: a bit slow

Sqlite
======

* part of cpython stdlib since 2.5

* we use Gerhard Haering's CTypes version

* works reasonably well after some fixes

Django
======

* we run unmodified Django 1.0

* only sqlite DB backend for now

http://www.djangoproject.com

http://code.djangoproject.com/wiki/DjangoAndPyPy

Pylons
======

* worked almost out of the box once eggs
  were working (1 day)

* no SQLAlchemy yet, obscure problems
  ahead

* unmodified passes all tests

* http://pylonshq.com/

Twisted & Nevow
===============

* twisted works (60/4500 tests failing)

* nevow works

* we don't support PyCrypto nor PyOpenSSL and we
  won't anytime soon (if nobody contributes CTypes or rpython
  versions)

* http://twistedmatrix.com/

Stackless
=========

* We support stackless

* tasklets, frame pickling, greenlets

* fully cross-platform

Other software
==============

* pure python should just work 

* BitTorrent

* PyPy translation toolchain

* py lib 

* sympy

* various smaller things, templating engines

Obscure details that people rely on
=======================================

* non-string keys in __dict__ of types

* exact naming of a list comprehension variable

* relying on untested and undocumented private stuff

* exact message matching in exception catching
  code

* refcounting details

Conclusion on Compatibility
============================

* lessons learned: There is no feature obscure enough for people
  not to rely on it. 

* pypy-c interpreter probably the most compatible to CPython 2.5

* main blocker for running apps will be missing external modules

Speed - comparison with CPython
===============================

* we're something between 0.8-4x slower than
  CPython on various benchmarks without JIT

* our JIT is a huge leap ahead

* pypy-c has fastest Interpreter startup

Speed - JIT generator
=====================

* 20-30x faster on small examples

* nice proof of concept

* a bit of time needed to speed up large python
  programs

* probably won't make it to 1.1

* completely separated from the interpreter

Memory - comparison with CPython
===================================

* PyPy has smaller Python objects

* user class instances often 50% of CPython size! 

* PyPy has pluggable Garbage Collection 

Threading / Stackless
===================================

* currently using GIL

* free threading? "it's work" 

* pypy-c has software threading / stackless

* no modifications to interpreter involved

Other backends
==============

* PyPy-CLI runs!  Still needs a bit more integration with .NET

* PyPy-JVM runs, little integration so far

* general speed improvements

* both backends are progressing - very slowly though

* contributors wanted!

pypy-c on small devices
===============================

- cross-compilation 

- startup time 

- security 

- RAM usage 

- share interpreter state across processes

- pypy approach a very good fit! 

1.1 release
===================================

- compatible to Python 2.5.2 

- well tested on win/linux 32 bit 

- running major packages unmodified 

- easy_install/distutils working 

- help e.g. by writing ctypes modules 

- hopefully released next month

Contact / Q&A 
==========================

holger krekel, Maciej Fijalkowski
at http://merlinux.eu

PyPy: http://codespeak.net/pypy

Blog: http://morepypy.blogspot.com

.. raw:: latex

    \begin{figure}
    \includegraphics[width=64px,height=64px]{merlinux-logo.jpg}
    \qquad
    \includegraphics[width=80px]{../../img/py-web.png}
    \end{figure}