Thursday, July 29, 2004

OSCON 2004 Day 3

The first day of the actual conference sessions went fantastically.

We started off with the opening keynote, from Tim O'Reilly, where he talked about what was on O'Reilly's radar. In addition to the "Internet Scale Applications" being the "Open Source Killer Apps" (like e-bay, google, amazon, etc), which he's been talking about for a while, he touched on a lot of interesting little bits. I particularly liked the part where he showed some of the data they mine from bookstores like Amazon that indicates what programming languages are currently selling the most books. The ones with declining market shares were in red, rising in green, and size of the language's square showed the size of the market share. Most stuff was either slightly green or red (Tim said it was the summer and all the hackers were on vacation), but there was this tiny little sliver of BRIGHT GREEN off in the corner, and that was Ruby. That was a breath of fresh air. He also mentioned zeroconf as being something to keep an eye on, which is rather close to my heart, and went into his standard thing about how what O'Reilly does is find the new and interesting things people are doing and put them into text form so people can read about them. Conferences like this one are where they find the new and interesting things.

Next up was The Semasiology of Open Source, by Robert Lefkowitz. He ran out of time part way through, but the talk itself was a fascinating look at the origins of some of the words that the open source world uses constantly. Quite interesting, with some nicely placed "The Princess Bride" clips for good measure, but if I try to explain it I just won't do it justice. All I wish is that he'd had time to finish up.

After the keynotes finished up I gave my first talk of the conference, "Using the Apache Portable Runtime in a Non-httpd Application", where I talked about the basics of APR and some of the issues we came across while using it to develop Subversion. I think the talk went well, and I was pleased to see Stas and Geoff of mod_perl fame sitting in the audience. These two guys have more APR experience than I'll ever have. I was able to raise at least one question in their minds though, and we've been hashing it out on the APR and HTTPD dev lists ever since. Overall I think the talk went well, and several people have since come up to me and thanked me, so I guess my first impression was right.

Next I stuck around in that same room for James Duncan's talk on "Enterprise Perl". He had a lot to say about how to make your code easier to understand and how that can help you build systems that are more reliable and debuggable. I was particularly amused by his definition of "Enterprise Code". He says it's "Code that doesn't make money itself, but that helps the company make money indirectly". At the end of his talk he noted that it was brought to you by 10 years of code that doesn't make money. I don't know if I agree with all his statements, but they all seem worth considering at the very least, and that's what's really important here.

Next up I hit Miguel de Icaza's talk on Mono 1.0. I probably could have skipped this one, as I knew most of what he was saying already since I follow the project reasonably closely. Unfortunately, he got cut off just as he was getting to the interesting technical bits at the end about where they're hoping to go in the future. Oh well, maybe next time...

After that I went to listen to Justin talk about deploying the Apache HTTPD server within the Apache Software Foundation. These guys get a scary amount of traffic, and they handle it all with minimal hardware and a volunteer staff, and it turns out that the load on their boxes is always ridiculously low, so they've figured some stuff out and are worth listening to. He also covered some of the details of how the Apache infrastructure team works with the version controls systems they provide and the email infrastructure they provide, since those are the other two main services the Apache projects rely on.

Up next was the Perl lightning talks, which were interesting, as expected. Despite technical difficulties (to be expected any time you have that many people and that many laptops) they got through everything and I learned a bit. I'm sure others have already blogged about the specifics, so I'll just leave it at that.

The last session of the day was about Groovy, the new Agile (perl/python/ruby like) programming language James Strachan (who went out drinking with us on monday) is building for the Java virtual machine. It's fascinating stuff, and is definately good to see, if I did any kind of real Java development I'd be all over this stuff.

Once the sessions wrapped up Greg Stein, Justin, and I hung around waiting for Fitz to finish his book signing. As usual when you hang around Greg you get this constant stream of interesting people who show up. Included in the stream this time was James Duncan Davidson (the original Ant guy), Brian Behelendorf (I must have spelled that wrong, but one of the head Collabnet guys and an original Apache hacker), Jason Hunter (Java Servlets guy), and a cluster of Gentoo Linux guys. Interesting conversation all around.

When Fitz finished up we headed out in search of food, and after much wandering around ended up in a not-so-great mexican place, with another large crowd of interesting people. Along with many of those already mentioned we had Dick Hardt (formerly of ActiveState) and John Viega (a security hacker now, who apparently was also one of the original authors of mailman, one of the big time mailing list managers).

After dinner some of us hopped in Dick's Hummer (yes, some of these people have way too much money) and drove over to the Stonehenge party. "Free as in Beer" was the order of the evening, and a good time was had by all. I spent some time chatting with Brian Behelendorf and Paul Hammant (who I met on monday) about version control type things, and I was quite glad that I got a chance to thank Brian for everything he's done to make Subversion a success. While he's not contributing code he is the man behind the fact that Collabnet is willing to pay people to hack on it, and for that I'll be eternally grateful. While the external community has contributed a TON to Subversion over the years without those core paid hackers we wouldn't have made it nearly as far or as fast as we have.

Around midnight Paul and I wandered back towards the hotel, talking about version control, why people use clearcase, the chances for perforce to survive in the next few years and his background and how he ended up at ThoughtWorks. All in all an evening well spent.