PyPy’s Approach to Implementing Dynamic Languages Using a Tracing JIT Compiler

Date

January 31, 2011

Speaker

Carl Friedrich Bolz

Overview

Dynamic languages are growing more popular, however, the restrictions of many of their implementations prevent them from realizing their full potential. Writing a dynamic language implementation that is fast as well as flexible and maintainable remains difficult. In this talk I want to analyze the reasons for why it is so hard to make dynamic languages fast and discuss how various implementation approaches try to address these problems.

In the second part of the talk I will give an introduction into the PyPy project and its approach to virtual machine construction. PyPy is an environment to implement dynamic languages way using a modular tracing JIT. It hopes to overcome the difficulties and make implementing fast dynamic languages significantly easier.

Speakers

Carl Friedrich Bolz

Carl Friedrich Bolz is a PhD researcher at the University of Düsseldorf, Germany. He is broadly interested in the efficient implementation of dynamic programming languages. In 2005 he started to contribute to the PyPy project and subsequently became one of the core developers. He has been involved in nearly all areas of PyPy development, particularly in the development of PyPy tracing JIT compiler.