Sun 29 Jul 2007
OSCON, part II
Posted at 09:26 +1000
(This a follow-up to this item, which may not have been syndicated to all the same places, since it isn't in exactly the same categories.)
Selected moments from the last couple of days of OSCON:
The Python BOF on Wednesday night. Excellently moderated by David Goodger. Probably around 20 or so people there. Many being fairly experienced Python users. A general consensus that selling Python to the PHB level of management is still senselessly hard. It's almost an argument for certification. All involved in the hiring process seemed to agree that wouldn't change the way they hired, though. Teaching Python to good people is easy and Python's the type of language where you can get a good feeling about an existing developer's work in a number of ways. A piece of paper won't change that. Other topics including code organisation, other languages used, promoting certain packages as reference implementations were all interesting. Over two hours of good conversation (and beers).
The Django BOF on the following night was well attended, too. Around 18 people, ranging from a few who were just starting out to some who were using it in high profile site. Conversation mostly driven by the more experienced users. I came away with a list of about 15 things that we can look at that I need to write up as an email to django-dev over the next day or two. Hallway conversation after the hour-long BOF went for another 90 minutes with a different group of people.
A talk on 37signals' experiences with using Amazon's S3 for mass storage was interesting. Mostly it was a litany of woes about "what can go wrong will", but it provided some very constructive lessons learnt about the extra work you have to do and expectations if you're going to hitch your business implementation to something like S3.
r0ml is a very entertaining speaker and I enjoy hearing him talk. Always something to learn about presentation. However, his point and generally excellent arguments are sometimes obscured by his entertainment style and this is a shame. It took a lot of work to sort wheat from chaff in his Preventing Code presentation.
Greg Kroah-Hartman's talk on the state of the Linux kernel in 2007 will be useful as data for the future. He's done a lot of crunching of numbers with regards to number of contributors (including nice chart), number of lines of code, proportion of change in various subsystems, and so forth. The simple conclusion is that anybody trying to maintain a parallel branch of the kernel is looking impossibility in the face. Pushing things into the mainline is the only way to go, because it's just too frantic. And yet, the system mostly works. The kernel remains healthy and solid.
Having contact with other Django developers and users during the day was very motivating. I got a lot of coding done during downtime between talks and in the evening back at the hotel. Some big ticket items, too (since the little ticket stuff tends to take care of itself).
The last few days have been a bit of a blur, so there might be an OSCON, part III post once my brain settles down and organises things.
Mostly recovering today and doing very little. Off to New York City tomorrow to enjoy ... I don't know ... stuff. Weather. Central Park. Manhattan Restaurants. Things like that.
Topics: software/django, conferences/OSCon, travel