I just uploaded the last of my pictures of ApacheCon US 2005. If I promised to e-mail you a picture, you are next...

Here's what's left of my thoughts on this year's con.

I like San Diego better than Las Vegas. There aren't as many things to do, but I never did any of those things anyway, and I did spent a lot more time (if that was possible) talking to people this year. The new production company, Full Circle Productions, did a great job. As far as I was concerned all the logistics were handled very efficiently.

I chaired a number of sessions, mostly of friends or people who were talking about something that interested me: Lenya+Jackrabbit, CalDAV, Web Services, and Ruby on Rails. It's hard for me to get excited about going to conference sessions on technical topics, because quite frequently I can visit the website, read the docs, and achieve the same level of understanding in approximately the same amount of time. Every once in a while, someone does stuff in a session that is hard to figure out from the docs, and these sessions are more worthwhile for me. I really liked Brian McCallister's Ruby on Rails talk. He used quicktime moves to playback executions of the various Ruby on Rails scripts, while he narrated what happened. Great style.

This year ApacheCon had a mini business track that ran through Monday. Monday ended up being a bad day for me because I set, but failed to turn on the alarm clock, thereby missing Cory Doctorow's keynote, which I was actually interested in seeing. I walked in a bit late to Simon Phipps' The Zen of Free, but I liked what I heard. I agree that open source governance is going to become a bigger issue in the future, and that lots of people who don't currently see the value of this will do so in the coming year. I need to get Simon's slides and have a good look at them to see what I missed. Jim Driscoll did a presentation called Open Source for Business and Profit. There were a couple of really good suggestions that Jim had for companies trying to climb the open source learning curve. I need to get a hold of his slides as well.

That leaves the keynotes. I missed Cory Doctorow's. I already blogged a bit about Tim Bray's, but here are a few more tidbits:. Good quote "I'm going to ssh into your internal Tivo". Tim was sitting in the back of my presentation - I was wondering what he was doing there, since my presentation is absolutely at the novice level. Apparently he was gathering fodder for his keynote, because he related several things about what happened during the session (two FOP committers met each other in real life during the Q&A). Tim commented that we (the Apache community) are a very polite community and that perhaps we were too polite and that we needed to wade into some of the various controversies roiling the web. I guess I need to make the web services barbs in my talk a bit sharper. His keynote was wide ranging, from an announcement, to serious talk about multithreaded hardware (he brought a Niagra for show and tell). He suggested Erlang as a language to look at for how to deal with concurrency problems, and suggested that Apache is going to run headlong into all of the problems related to highly concurrent hardware. He also ventured briefly into Beyond Java territory, but alas my notes say no more than that. He concluded his keynote with a "thanks" (he was one of the users of the stream of patches that became Apache httpd), and an admonition: "Don't screw up". Indeed.

There was an Oracle "keynote" before the lightning talks. It was pretty bad. It was basically a product pitch, along with an announcement of a future donation of Oracle's ADF to the MyFaces project. If you are coming to ApacheCon to get the Apache community on your side and interested in your proposed project, this is not the way to do it.

Sadly, the Intel keynote wasn't much better. The speaker seemed to be in awe that he was speaking at ApacheCon. I expect a keynote on Open Source from the world leader in microprocessors to be a bit better than "wow, I am so impressed" and "please don't me mad at me if I got it wrong". If you want to become part of the community, take some time to understand what the community is about, and find a way to show us how your issues are our issues. More multithreaded/multi-cored hardware, some open source BIOS, and contributions of security code for Harmony.

I fell asleep for the middle section of Jaron Lanier's keynote. But it wasn't the content that put me to sleep, it was just 5 days of full steam ahead catching up with a guy who was sitting on the floor with his back against a wall. Lanier started out on his disagreement with RMS on building a free operating system. Lanier didn't think that cloning Unix was that interesting -- I think that he was right. From there he launched into a discussion about the Singularity. Lanier doubts that such a thing will actually occur, and points out that the bigger software projects get, the more they fail. He went on to talk about how software is brittle -- it can't improve incrementally, and it accumulates in layers. The problem with this is that ideas, critical ideas, get lock in along with the software design. And because it is hard to go down into the layers and get them out, those ideas get locked into our ideas of how computers ought to be. He cited the example of how there used to be a debate about whether the notion of a file was a good thing or a bad thing. He felt that hope for the future of software was going to come from "the other side of computer science" - robotics, simulation, etc. In order to interact with reality, you can't use protocols. Then there was an interlude where I fell asleep. When I woke up, he was just starting into a section on "What might a server of the future look like?". Lanier believes that computer interfaces will become more intimate (he is the VR guy after all), and that this will drastically impact existing servers. All of a sudden, timing becomes primary, because the server must be able to keep up to maintain the intimacy of the interface. He believes that the architecture of the net itself will have to change in order to accomodate the impact speed of light realities on these interfaces and the servers that provide them. He (third time's the charm) also brought up the problem of programming multicore processors, and mentioned partitioning applications across cores as a possible solution (one core for the AI, one core for the rendering, etc). The last section of his talk was about the style of the open source community. He said that the open source community naturally gravitates towards an additive style (here's a patch that adds x) versus a subtractive style. It's easy to keep adding stuff (ease is relative), but when you are dealing with a user interface, you sometimes need to subtract in order to get to a good interface. Lanier suggested that we need a more vigorous form of subtractive culture in open source. He hearkened back to the earlier portion of his talk where he talked about brittle software vs natural selection/evolution, and said that the key thing in evolution is selection. Selection subtracts, and it works by death or broken hearts (someone didn't get chosen as a mate). He would up with some jokes and stuff. It wasn't the most organized keynote that I've ever heard, but it was definitely the most thought provoking of the whole conference. I hope I get the chance to hear him again when I'm not bushwhacked by 5 days of ApacheCon.