PyCon 2008: the conference

A word of advice when attending PyCon: try not to show up ill.

I arrived on the conference Wednesday afternoon since my roommate had the room booked starting then and to get over the jet lag since my talk was Friday morning. But I came to PyCon with a cold and it drained me. Nothing much happened Wednesday except dinner with some people.

Thursday was the tutorials. I was not giving nor attending any, so I spent the day saying hi to people I know as they showed up at the hotel along with trying to get work on for my AOSD demo (I will give a complete blog post on my work once AOSD is over). It felt rather odd to be doing Java work at PyCon. =)

Friday was when the conference really started. The diamond keynote by White Oak Technologies was entertaining (i.g., I disagree with some people who have blogged that the diamond keynotes by sponsors was a bad idea). Guido's talk was not overly informative for me, but it is still interesting to see how the audience responds to what is going on in Py3kLand. It was also interesting to hear Guido say that I am probably regretting signing to be the lead on the stdlib reorg. Guido rarely mentions people by name when giving a talk, instead treating python-dev as a large group and all things coming out of it a group effort (or at least that is what it seems like; maybe he just doesn't remember who did what). When he bothers to mention someone by name I tend to take notice.  So having him publicly point out the pain I am dealing with was nice. =)

I gave my talk later that day. Having my earlier version of my talk flop at VanPyZ, I restructured it to be much more visual, using flowcharts to get key points across. I didn't practice this new version, and manager to finish in 17 minutes. That left me roughly 12 minutes to be peppered with questions. A few were clarification questions, but most were very specific problems people had encountered with import. I tried my best to answer them on my feet. I also got tracked down multiple times throughout the conference with more questions. Various people told me the talk went well, but I always worry I went too fast, didn't state things simply enough, etc.

The rest of Friday, for me, was PSF-related (my cold did not make it easy to stay focused for very long). The general members meeting happened and I got re-elected to the board. The board also had a meeting where I suggested that we try to pay to have more sprints occur for the core team so we can be productive more than just once a year at PyCon, which was well received. We also had an amazing dinner later where I had this huge hunk of prime rib which was amazing. After shocking people I finished my dinner I ordered dessert which turned out to be massive as well (there will be, I'm sure, photos of my and that dessert floating around on the Net once some board members upload their photos). Luckily almost everyone at the table helped me eat it.

In terms of talks on Friday I remember attending Adrian's "The State of Django" talk. It was fun to sit next to Jacob while Adrian was presenting to get insider info on various things that were going on. And congrats to the Django guys for announcing there movement forward on starting a foundation!

I also attended Friday's sprints. Yes, the sponsor sprints were a bad idea, but they had worked out in 2007. Anyway, they won't happen again. I remember Jerry Seutter's lightning talk the most from that day since it was extremely entertaining and admittedly written thanks to sleep deprivation.

Saturday, thanks to my cold, was a day to sleep in, so I missed Van's keynote (which I wanted to attend, darn it). I made Richard's talk on Pyglet, Maciej's talk on PyPy, and Michael's talk on Python and Silverlight. All three were interesting.

Also hit up the lightning talks. I decided to start a PyCon tradition for myself where every year I will give a lightning talk bitching about some facet of Python I don't like. Last year it was on tuple parameters and led to PEP 3113. This year I complained about various stuff related to import. I am hopeful that in Py3K I can get rid of implicit importers (e.g., built-in imports, etc.) and put them directly on sys.meta_path. This would allow for having an importer that deals with sys.path be put on sys.meta_path that treated sys.path as a fallback mechanism, removing sys.path_hooks and friends and subsume them into the individual importers. This would tweak import semantics by having how a module is written (extension module vs. bytecode) take precedence over location on sys.path, but if you have multiple modules with the same name on sys.path you are already asking for trouble. Lastly, a new function that makes it easier to directly import a module will most likely come into existence.

On Sunday I attended Mark Hammond's keynote on Python and Mozilla. It was interesting to hear him discuss the community problems that PyXPCOM has since the Python community thinks it is a Mozilla project and Mozilla thinks it's a Python project, and so no one claims it. I know I always thought of it as a PyXPCOM project.

I also attended Jim Hugunin's talk on IronPython. As always things are moving forward. They showed Django up and running using MS SQLServer running as the backend. I also hit up the lightning talks which were as good as Friday's.

Then my block of content came up. I gave my intro to sprinting while only lasted like 10 minutes. There was one silly picture early on in the talk that got a bunch of good laughs and I kept getting comments on throughout the sprints. I then led the sprint panel afterwards. I honestly think the panel was unneeded; there was only like three or four questions from the audience. I believe most people who showed up knew what sprints were and so they didn't really have any questions.

After that I gave my tutorial on core development. Actually met some people who had already contributed to Python while I was presenting (which I did from a chair sitting in front of everyone). People seemed to get something out of it which was good.

And that was the conference! I will do a separate post on the sprints. Overall I was pleased once again with PyCon. I was constantly blown away when I realized that there were over 1040 people at the conference. I remember PyCon 2003 when it was only about 250 people. The conference has finally reached a size where I actually went a full day without seeing some friends. It's definitely great to see Python grow so much and yet still have a community of very friendly people exist around it.