Clearly, PyPy is more than a just faster alternative to the standard Python implementation CPython with a JIT and an experimental Software transactional memory implementation. Some people even use it to implement their own programming languages.
We took the opportunity to catch the two core members Armin and Maciej of the PyPy project at a Sprint in Leipzig. They talk about the project, what they achieved so far, what's boiling at the moment, and what else you can achieve with the RPython toolchain.
You will have noticed it by now, we fell back to the least common denominator English as the interview language. We applolgise in advance for the restricted recording environment.
As a side note, this podcast now is also available via bittorrent from Bitlove.org.
Shownotes:
00:00:45 Intro of our special guests Armin and Maciej
00:04:33 CRE088 Python und PyPy
00:07:15 Speed gain and benchmarking
00:08:40 Codespeed / speed.pypy.org
00:09:15 PyPy project, contributing
00:16:13 Development workflow
00:20:25 PyPy in detail: the components and features
00:24:00 PyPy can be faster than C ;-)
00:26:15 What happens when you execute pypy or the Toolchain
00:30:27 A little bit about the JIT
00:33:50 RAM consumption
00:43:00 Sandboxing
00:45:33 Software and Hardware Transactional Memory (STM and HTM)/ Automatic Mutual Exclusion
01:01:17 What is the Global Interpreter Lock and Stackless?
01:05:50 How to support PyPy or sub projects?
01:08:10 Who uses PyPy?
01:08:10 Who uses PyPy? Which Python projects support PyPy?
01:11:10 How to file bug reports?
01:14:36 Outlook.