.. include:: beamerdefs.txt

=============================================================================
Writing Interpreters for Dynamic Languages Using PyPy (and Getting Them Fast)
=============================================================================

Motivation
================

- good VMs for dynamic languages are very hard
- when writing it is very hard to reconcile

  - performance
  - simplicity, maintainability
  - flexibility, features

- this is particularly true in an Open Source / research context

Python Case
============

- **CPython** is a simple maintainable but slow VM, using a simple bytecode
  interpreter
- **Stackless** is a more featureful fork of CPython that was never merged for maintainability reasons
- **Psyco** is a very complex JIT compiler for Python, which speeds up Python a lot

  
Motivation (2)
================

fixing of early design decisions:

- when writing a VM in C, some things have to be decided up front
- examples are things like memory model, GC, threading model, etc.
- decisions manifest throughout the source code
- extremely hard to change later

|pause|
|alert<| Python case |>|

- reference counting
- OS threads

|end_alert|

PyPy's approach to VM construction
==================================

Goal: achieve flexibility, simplicity and performance together

- Approach: auto-generate VMs from high-level descriptions
  of the language
- high-level description: an interpreter written in a high-level
  language
- ... which we translate (i.e. compile) to VMs running in
  various target environments, like C/Posix, CLR, JVM

|pause|

- the high-level language that is used to implement the interpreters is called *RPython*

PyPy's approach to VM construction
==================================

.. image:: dynlang.png
   :scale: 50

Receipe for a PyPy-based VM
============================

 #. write an interpreter for your language in RPython
 #. test it in pure Python until it works and is bug-free
 #. try to compile it to C and fix the type-bugs that this finds
 #. apply the JIT generator (optional and experimental step)

What is RPython
===============

A subset of the Python language on which type inference can be performed

- single inheritance with mixin support
- most of Python's builtin data structures are supported
- no type mixing: ``l = [1, "string"]`` does not work
- don't think about it too hard, it feels a lot like Python still