Friday, December 23, 2011


The Talk schedule has been published and I'm on it. For which I'm very greatful - I saw the full list of ~400 submissions and there were many I'd like to see which didn't make the cut. [To all those people I'll let you in on a secret: submit the same thing at a smaller conference. They have 80%+ acceptance rates - basically they just toss out the crazies and let everyone else speak. PyCon was like that not too long ago. So go and speak wherever they will have you and get some practice: my first PyCon talk never would have been accepted under today's standards.]

Name Dropping

Normally I'm a Hallway Track kind of guy but there are some talks from friends I kinda want to see as it happens as an event: Larry Hastings, Doug Hellman, Jeff Rush, Benjamin Peterson (is he old enough to drink yet?), David Mertz, Moshe Zadka (long time/first time?), Ned Batchelder, Alex Martelli, Barry Warsaw, Dino Viehland, Maciej/Gaynor/Rigo (pypy), Armin Ronacher, Glyph (I'll tell you his real name in trade for a beer), Raymond Hettinger, Gary Bernhardt, and Jim Baker.

And that's just the list of speakers I know will be good, there are many more that look good on paper that I want to see.

You'd better nail it kid

There are some perennial talk topics because it is always nice to have a fresh take on important things. That said, being the second (or third, or fourth) person to retread the same ground is difficult.

Larry Hastings is treading old ground with Stepping through CPython, a tour of what really happens when your python code runs. I did it in 2006 as "Writing your own python types in C" and Ned Batchelder did it much better in 2009 as "A Whirlwind Excursion Through Python in C". I'm looking forward to Larry's take.

Getting the most out of Python imports by Eric Snow. Note that he speaking about imports and his name isn't Brett Cannon.

The Magic of Metaprogramming by Jeff Rush and Metaclasses by Luke Sneeringer. Not just well tread ground, but ground always loaded with fresh landmines.

Decorators and Context Managers by Dave Brondsema. I'm very much looking forward to this one, as those things have been my topics in the last few years. I haven't met Dave but I imagine he has giant brass balls.

And the other thing

Raymond Hettinger will be presenting The Art of Subclassing and I will be presenting Stop Writing Classes. Raymond and I have agreed on pretty much everything in the decade I've been involved with python (he was already a commiter when I showed up*). I doubt we are in disagreement about classes or subclassing, rather we are giving talks about two different topics. His is "here is how you do hard things well" and mine is "don't do hard things, they are hard and you'll probably fuck it up". Two weeks ago I wrote a caching class that used __new__ and this past week I refactored it to not do that; the code was half the length and achieved the same result. My only question to the team was "why did no one yell at me two weeks ago?"

* My favorite PyCon story is my first one: 2003 in DC and I got to meet everyone that was previously just names on a mailing list. Hettinger was working the registration desk and I was like "thats neat" and when I stepped out into the smoker's lounge I introduced myself to the three hangers about. "Hi, I'm Jack Diederich" was followed by "Hi, I'm (Alex Martelli|Tim Peters|Christian Tismer)". That was awesome.


Larry Hastings said...

Glad to make your list!

Your talk was before my time, and Ned's talk (and sounds like yours too) had a slightly different focus. My goal is to get people conversant with CPython's architecture so they can become core developers. Obviously it's all the same API, but I plan to talk about the implementation too (e.g. how the small block allocator works) and about exotic topics like how Python bootstraps itself.

price per head call center said...

This is really satisfied by the nice services in this blog that to really providing the wonderful info is visible in this blog

Paddy3118 said...

Saw the video and agreed with you, classes can obscure.

The only thing I would add is maybe a positive mention of namedtuples and how their use can be succinct and clear.