Monday, July 26, 2010

EuroPython 2010 report

So, EuroPython 2010 is over, I am flying home and it's time to write a report about the conference from the PyPy point of view.

As usual, the conference was very interesting and went very well. The quality of the talks I attended to was high on average and most importantly I could meet a lot of interesting people to discuss various things.

On the first day, Armin, Amaury and I presented the usual PyPy status talk (here are the slides): the talk is an extended version of the one that I and Armin presented at Pycon Italia in May and is divided in three parts: first I talked about the current status of the project, what is the content of the recent 1.2 and 1.3 releases and showed a demo of a simple Django application that renders a Mandelbrot fractal and is measurably faster on PyPy than on CPython. In the second part of the talk, Armin gave an introduction about the ideas that stand behind the JIT. Finally, in the third part Amaury explained how the new cpyext module lets PyPy to compile and load existing CPython extensions written in C.

I think that the talk was well received: the only drawback is that there was no time to answer questions at the end of the presentation. However, we received a lot of "offline" questions after the talk finished and thorough the whole conference: it is always great to see that people are interested in our work, and I'd like to thank everybody for the feedback that they gave to us.

PyPy was also mentioned in the interesting Mark Shannon's talk, where he compared the optimization techniques used by PyPy, Unladen Swallow and HotPy, which is Mark's own PhD project. Moreover, Henrik Vendelbo gave a talk about how to tweak PyPy to produce a standalone executable which embeds a whole python application to make deployment easier, while Andrew Francis explained his implementation of the Go select statement based on the stackless.py module implemented in PyPy. Personally, I am glad to see that people start to think of PyPy as a useful starting point to experiment with new features and use cases that we did not think about: after all, one of PyPy explicit goals is to be "flexible and easy to experiment with".

After the conference there were the usual post EuroPython sprints: this year we had not planned a PyPy sprint, but some people showed interest in it and since Armin and I happened to be still around the day after the conference, we decided to do a mini 1-day sprint, with 6 or 7 people present. Since there were only two core developers it was impossible to use our usual pairing scheme, in which every newcomer pairs with someone who is experienced with the source code to gain knowledge of it. However, I think it was still a successful day of work, and we managed to fix a couple of bugs that was standing in our issue tracker. Again, I'd like to thank all the people that came and worked with us during the sprint.

In conclusion I really enjoyed the EuroPython 2010 experience: the fact that I managed to find a place in Birmingham where to eat a good Italian-style "gelato" helped a lot :-).

2 comments:

  1. Finding gelato hopefully won't be a problem at next year's EuroPython. ;-)

    ReplyDelete

See also PyPy's IRC channel: #pypy at freenode.net, or the pypy-dev mailing list.
If the blog post is old, it is pointless to ask questions here about it---you're unlikely to get an answer.