I didn’t find the quality as good today. The keynote was bad and I picked a couple sessions that didn’t hit the spot. Note to O’Reilly: if we are going to come in at 8:30 AM, we need to hear things we haven’t already heard and they need to be from people we will listen to. Gary Lang from AutoDesk was a joke. He was a suit that didn’t have any understanding of open source and he made AutoDesk look bad.

Sessions I Liked

• Subversion Best Practices – Ben Collins-Sussman and Brian Fitzpatrick (both now at Google) gave a great talk on Subversion. These two literally wrote the book on subversion. The did the right thing and didn’t give a talk on what subversion is. Instead, they answered many questions on how to use it properly. They covered structuring the repository (one is better than many), which server to use (apache is most flexible, others have their place), permission models (less is better), and many other smaller tips. I recommend finding the slides at the OSCON presentations page and reading them.
• Google Code Hosting – The big news was Google has released a new code hosting service. It is aimed at the heart of SourceForge. And I think it will have a big impact on it. Google’s code hosting has a subversion repository with 100MB of storage, an issue tracker reminiscent of GMail, and a small home page. One other nice thing is that they have you pick 1 license from seven different options. Greg Stein of Google said that they are taking a stand and pushing projects to pick one and only one license. It’s built from subversion running on something called BigTable which is a Google proprietary storage back end. Disappointing that they didn’t open source BigTable or the hosting code. Now I know why Ben and Brian are at Google. I’ll post more on Google code sharing later. I have a project in mind that I may put in there. I think it has great potential.
• Google Web Kit – This was new to me. Basically, you write your Ajax code in Java and use a Java to JavaScript compiler to generator your JavaScript. You can tie it into the Eclipse debugger (if you aren’t on a Mac – lame). There is a good community of widget developers forming already. I’m not a UI guy but I thought it looked useful. My co-worker Scott Haug has been tracking it since it was originally released. He said it doesn’t fit what we are doing. I’ll have to have him give me some more info on it when I get back.
• Ubuntu and Community Building – Jeff Waugh did a nice presentation on why Ubuntu’s community has taken off so rapidly. They made it a very clear goal early on and it shows. Like anything that becomes a big success, they aimed big and worked hard to reach it. One interesting question asked of him during the Q&A regarded the distribution of non-free drivers from NVidia and ATI. Personally, I don’t have a big problem with this. I care more about getting my hardware running. Some care more about the origin of the code. Jeff had a nice answer. They did distribute the non-free drivers to create an acceptable out of the box experience for Ubuntu. But they also put in a hardware report option that people can send their hardware config to Ubuntu. Jeff has been able to use this to validate things like how many people are able to successfully run the Intel graphics chips (which have free drivers). Those numbers give him the leverage to shame ATI/nVidia into doing the right thing. It’s a good approach – the hardware works and pressure is put on the hardware manufacturers to open up the API’s to their hardware.

Sessions I Didn’t LIke

• Python 3000 – Guido van Rossum gave an overview of the next version of Python. Unfortunately, Guido wasn’t feeling too good. And I was more than a little underwhelmed by what is going in. In fairness, I don’t have any major issues with Python (although I’m not a heavy user). It does seem a little weird that moving the strings to UTF-8 is a big deal in 2006. Good this is happening but seems like a given these days.
• MySQL Replication – First off, it’s pretty boring to hear about yet another catch-up feature in MySQL. A big feature for 5.0/5.1 is that they are doing row level transaction log replication. About time. We did it ages ago in SQL Server and Oracle has as well. I get that MySQL is “good enough” most of the time and is relatively cheap. But I really wish the darling of the open source db world was Postgres and not MySQL. At least it feels like the Postgres guys come up with new ideas.

One more half day and then back home. My wife has been down here for the past two days as well so we’ve been able to enjoy Portland in the evenings and catch up with some friends. I’ve enjoyed the conference. It’s been more fun to be here and just learn than working like I did last year.