Kevin's Weblog

I couldn’t take detailed notes this year on account of a sprained wrist, but here’s the highlights of my week.

I spent most of Monday and Tuesday working on my talk. Tuesday afternoon, I bounced back and forth between People for Geeks and Francesco’s Erlang Tutorial. Nothing new for me in the Erlang talk, but I wanted to get an idea of what he was covering. There was a lot packed into 3 hours, but people I talked to said positive things — that they learned a lot about the language. Over in “People for Geeks”, Andy Lester had quite a bit to say about working with your manager which I considered questionable. For example, “don’t go to your manager with problems” (I thought a manager’s job was to clear obstacles) and “your job is to make your manager look good” (I wonder what our shareholders and upper management have to say about that one). On the other hand, pointing out that your manager is probably not actually incompetent or malicious is valid, although it’s a little sad that we have to point that out.

Tuesday evening was the opening extravaganza. I was kinda bored by Mark Shuttleworth’s talk, but r0ml and Damian Conway were excellent as usual. Although, knowing enough physics to be bothered by the inaccuracies in Conway’s talk was a little annoying.

Wednesday I skipped the keynotes in favor of a proper breakfast and another practice run before giving my talk. Seemed to go well — good attendance, questions that indicated people were interested in the subject matter. Bummer that they scheduled both Erlang-related talks at the same time, though. A talk about Hypertable was interesting, although a few too many questions were answered by “we’re still working on that”. Wrapped up the day with Paul Fenwick’s An Illustrated History of Failure, a fun overview of some of the more spectacular (mostly software) failures the world has seen. In the evening, went by the MySQL / Zend / Sun party, but spent most of the time worrying that drunk people on tricycles were going to run into me and complicate my injuries. Also, giant rabbits are creepy.

Thursday. Nat’s keynote on teaching kids to program was inspirational. I downloaded the programs he recommended to try out with my daughter. Processing Large Data with Hadoop and EC2 was quite interesting. The New York Times used Hadoop and EC2 to do a pile of image processing related to releasing the archives of the first 100 or so years of the paper onto the web. Tim Bunce talked about Ultimate Perl Code Profiling using Devel::NYTProf v2. To quote an audience member, “I have only one question: why are you so awesome?”. I also went to a talk about Meebo’s Interview Process which was pretty interesting. I’m not sure that I picked up anything new, but it did validate a number of opinions I’ve formed about what works and what doesn’t over the years. In particular, having candidates spend a couple hours actually doing the sort of work they will on the job is a really valuable technique. She’s stressed that you have to ask people to do a task that people who don’t do that job would actually fail. I’d add that your interview process should probably be rigorous enough that some of your current employees who are doing that job would fail it.

Thursday evening I decided to go home for a proper dinner, then came back for Beerforge (kinda disappointingly sparsely attended) and the O’Reilly speakers’ party (full of interesting people, unsurprisingly).

On Friday, Tim Bray’s keynote was an amusingly frantic discussion of language trends, with a nice shout-out given to Erlang, and a general leaning towards functional languages as a promising developing direction. The Twilight Perl was a little less scary than average Conway, because he limited himself to no source filters or even outside modules for the most part. As a result, most of the tricks weren’t too difficult to figure out if you’re familiar with all the corners of the base language. But, it was still quite entertaining. After his talk, I bailed on the closing keynotes (a repeat of “An Illustrated History of Failure” for some reason) in favor of lunch and trying to replace my missing phone (lost somewhere Weds evening).

Two fan-boy moments: Spent a few minutes talking with Larry Wall about Erlang, and asked Tim Bray for advice about hats (specifically traveling with them).

Overall, it was a good week as usual. Once again, I was struck by how much I felt that living in town actually detracted from the conference experience. If I submit a talk for next year, I’ll make sure to actually start writing it more than a week in advance; that was stress I didn’t need.