Pycon 2014 was held in Montréal from 11-13 April.
Since I stopped going to European python conference, I've been to 7 on my own continent - Chicago twice, Atlanta twice, Santa Clara, Toronto, and now Montréal. This is the third time I've gone by car - which has the main benefit of not dealing with narrow seats and US airports, actually having my own car in the destination city hasn't been all that useful (except for Santa Clara, where you need a car just to get to the nearest McDonalds - but that trip also included a week of photo-tourism and 100+ miles of driving in it's own right.)
I originally planned a return to Mt. St. Bruno for hiking and birdwatching, but the weather and circumstances didn't cooperate (also, while Boston was just starting to stumble into spring, Montréal was really still convincingly in "late winter".)
Entertainingly the border crossing, always straightforward, was even
more direct than usual with an "Oh, you're going to that tech
conference?" :-)
The enormous conference center was easy to find by car, and Montréal is one of those cities with enough parking. Both conference hotels had underground access, and were about the same distance to the actual conference session rooms (not at all obvious in advance; the Hyatt looked to be farther on the map, but the Westin was at the far end of the Palais building and you had to walk the entire length of it to reach the actual Pycon rooms.) Once you got to the conference, it turned out to be stacked vertically - aside from escalator congestion, it really wasn't much effort to change tracks/rooms between talks, and the Vendor Room was nearby without actually interfering.
Also, the whole place was brightly colored - variegated windows, pink tree sculptures, color schemes to help you remember what floor you were on...
Didn't really dive in to nearby restaurants, but "Steak Frites" right near the weston, "Eggspectations" attached to the Hyatt, and a Szechuan place about 10 blocks away satisfied my need to demonstrate I could actually get around a real city and not just fall back on conference and hotel food.
Most useful gadget: Galaxy Note 3 with Guidebook installed, made it easy to pick and choose talks and set reminders in any free moment. The built-in regrettably-proprietary Action Note was a fine "pad of stickies" for scribbling down details, though I made more use of the Canon S100 to grab pictures of slides for reminders.
For complicated reasons I also had a chromebook and laptop with me; the instant-on of the chromebook was useful once or twice (though I could probably have used the Note for everything I did with the Chromebook, especially since I had the bluetooth keyboard on me) but I certainly wouldn't have missed it (and didn't use it during talks at all.) Overall, my gadgets helped me engage, instead of distracting me, more than at any previous talk - even letting me go so far as to turn a passing mention of "that book my spouse edited" into an amazon kindle-purchase and immediate read.
After the first day I didn't carry the big camera, either; having my hands free to grab the phone was much more useful, and there wasn't much city wildlife to shoot anyway. The full suite of pockets and removable sleeves made the SCOTTEVEST the second most useful gadget :-)
A head cold (picked up back home, not at the con itself) tired me out and caused me to miss the early keynotes (until I discovered that you could just buy real sudafed OTC, no id, which meant I was functional but twitchy for VanL and Guido's keynote talks; the high point was the long list of Raymond Hettinger's contributions as part of his Lifetime Achievement award. (Fortunately the excellent http://pyvideo.org coverage means I can go back and catch up on the other 120+ talks I missed, as well as linking to the ones I did attend and can recommend.)
Diving into Open Data with IPython Notebook & Pandas was a great introduction to a topic many of my coworkers are already well-versed in, namely, what's a good "pocket calculator with statistics" for moderately large data sets. The worked example from local Montréal cycling data was engaging, the speaker very enthusiastic (and rigorous enough to point out where bits of the talk were good presentation but not bad data science - not that it necessarily sunk in with the beginners, but it nicely deflected arguments from experienced analysts who might only have been there to vet the talk for recommending to others...)
iPython Notebook has been an awesome new thing for several years running, and it's really taken off over the last two or three years (a no-longer-surprising number of talks are given directly from it, these days - much more robust than google docs, since it's as local as you want it to be, as well as being more engagingly "live".)
Pandas has been picking up speed in just the last year - it's basically a python version of R's "data frame" but that still gives you a lot more leverage as a user (easier to explore available analysis methods, broad choice of plotting/graphing tools -- I'd never heard the name "violin plot" before, now I can easily just try it out on some data.)
Hello Physical World: A Crash Course on the Internet of Things showed how you can take a little bit of the technologies everyone is talking about (3d printing, arduino, opencv, zigbee) and build something real (in this case, a Skinner Box with a web interface and database for logging and analysis), without having to go deeply into any of the technologies (instead, you get to stay focussed on the problem you're trying to solve.)
Discovering Python was more of an engaging talk than a practical one, about a "desert island" coding experience involving the discovery phase of a patent suit, a vault with a Windows box, Python, and over a terabyte of source code. Not necessarily an inspiration for a career choice (though see Van Lindberg's talks from previous years on doing prior art analysis with Python), but as an opportunity to deeply engage in a problem, it's hard to top.
Granted I could just go back and rewatch his Idiomatic Python talk from last year, it was that good, but I didn't run across anything in the space of details like why adding/popping the end of the list instead of the start is much faster...
As far as I could tell, all lightning talks were prepared and selected
in advance. While this seems to have improved the quality overall, it
doesn't encourage a culture of "be brave and get up there and say
something" and I think there should be some space for that, even
given the size of the audience - right now they mostly looked like
practice runs for future full-length talks, which is also a good
thing to have, but I wanted to see a little more "inspired by one of
the talks on Friday, I Made A Thing on Saturday" too. (I did submit a
lightning talk that was more on the practice-run/is-there-interest
side of the scale, but in hindsight it isn't actually the kind of talk
I wanted to see so I actually have no complaints about not
presenting it :-) though I really do need to get around to
finishing it soon.)
I'll certainly be back, simply because it's the nearest pycon in years. I'll stick with the lunch-with-strangers/dinner-with-friends approach, that worked well, and I'll probably give up entirely on carrying either the laptop or big camera for the actual conference days and save that for surrounding vacation time.
The one thing I'll change is to try and make it easier to blog between (not during) sessions, and in evenings; getting this writeup done almost a week after the conference is less useful than it could be, though it does allow for more editing and link-collection.