[Dev] PyCon in 6 paragraphs
jeffrey at osafoundation.org
Wed Mar 1 16:50:46 PST 2006
PyCon 2006 was fun for me, just as the previous two years were. Moving
the conference to Dallas is a bit of a bummer from my perspective, I
prefer D.C., but the conference center in D.C. was too small.
The keynotes this year weren't very interesting, Guido's annual address
excepted, of course. There were a fair number of notable talks, though.
I thought all of the OSAF talks went well, Grant's talk on zanshin was
hilarious, and Brian's had examples of i18n issues (like graphs) I
hadn't ever thought about before. My talk seemed to be reasonably well
received, but I was so nervous that I went through my slides 7-8 minutes
faster than when I'd practiced in private, so my talk actually ended
early, a rare feat.
The questions at the Chandler BoF session seemed pretty on target, I had
the feeling that we'd reached a significant milestone in terms of
dissipating vapor and offering something geeks could believe might
really help them.
The pysense talk, about robots and movement modeling, was really
entertaining. It didn't bring me any closer to making myself a robot,
but seeing data analysis of a human wearing a camera and position
sensors was exciting. Apparently some of the folks at MIT's robotics
lab are using Python (with a fair amount of C++) to talk to robots.
The IronPython talk held promise, but unfortunately I'm wasn't familiar
enough with the CLR to really understand the code examples. It's clear
that IronPython is well on its way to being a really useful tool for
people to use the CLR.
I keep going to PyPy talks in the hopes that I'll understand what Python
running in Python is likely to offer us, this year's talk didn't help me
any more on that. PJE's explanation that a JIT compiler is actually
being worked on and may one day give near-CPython speed was exciting,
The talk on Nabu, a system for adding meta-data about anything (examples
focused on PIM data, though) in ReStructured Text and keeping a database
of the text and metadata was fascinating. I keep lots of text files
lying around and I like having prior art for when Chandler starts
working on marking up metadata in text (like, say, in a note or the body
of an email).
In previous years, sprints have happened before PyCon proper. I really
prefer that, it gives me a chance to build relationships with people
before the conference. On the plus side, I think having the sprints
after the talks helped vobject a lot. vobject is relatively simple, and
I think a lot of people would prefer to help out on fairly small projects.
There were four or five vobject sprinters, and they were really helpful.
I had my hands full answering questions, but without having me write
much myself, we got:
- improved directory structure
- fixed problems with running tests
- added photo support for vcard, including working around non-standard
format of vcards exported by Apple's Address Book
- Added vcard tests and documentation
- Added hCalendar serialization, although that code is in temporary
limbo while the Microsoft employee who wrote it makes sure no one's
going to get their knickers in a twist
Notes for next time
- Bring paper Contributor License Agreements next time, have people sign
them before they start coding
- Bring sweaters and long underwear next time I have to spend time
inside a hotel in Texas, gratuitous winter air-conditioning seems to
be the norm
More information about the Dev