So I didn’t have time to journal about the 2nd and 3rd days of the conference since there was so much good stuff going on. I’m on the plane back to Chicago now so I’ve got a few moments to jot down some notes about those days and some general thoughts about the conference.

To be honest the 2nd and 3rd days kind of blur together for me because I really didn’t get much sleep between them. I was pretty much blown away by the variety and quality of the presentations. Thom provided a detailed look at how he builds nimble, high-powered applications using short n’sweet python code on a beowulf cluster using techniques like map-reduce.

While they did separate talks on different topics I found some common strands between Devon Smith’s talk about metadata processing and Rob Sanderson’s talk about indexing in Cheshire3. Both of them had interesting workflows which they illustrated with neat diagrams which I should be able to link to from here soon. It wasn’t UML or anything boring like that. Rob’s illustration was more an overlayed animation over a bunch of slides showing the full lifecycle of a document being busted apart, indexed, a query coming in, triggering retrieval and then reconstitution of the document. Devon used interesting shaped objects to represent components in his metadata management framework. It was so much more fun than a dry description of what the software was doing, and really evoked what’s so much fun about building software–metaphor creation and architecture. Similarly Colleen Whitney of the California Digital Library had some really neat ways of visualizing search results which I wish I could link to as well.

Ryan Chute’s talk about the aDORe archiving framework from Los Alamos was interesting, but it largely seemed like a verbalization of the series of articles about aDORe that have been published. Don’t get me wrong, it’s fascinating stuff, and perhaps I just had super-high expectations–but I was hoping to hear more details of how they are actually using the aDORe framework at Los Alamos. It was good to hear Aaron Krowne talk about his experiments with quality metrics at Emory–especially after hearing a bit about it months ago in IRC. It turns out he was able to layer his new metrics over lucene without having to dive into the lucene code itself. I’m looking forward to seeing the code once it is released. I knew Aaron was a smart dude from talking to him IRC, but was surprised to see he is a confident and articulate public speaker as well.

Of course Roy Tennant is so at home at public speaking he was probably the only person that could easily tackle the “future of code4lib” in a presentation. He talked for 20 minutes about a variety of options that could be in the cards to make code4lib into a more formal organization; and then afterwards he did a breakout session on the topic. Unfortunately I wasn’t able to attend this because I was sitting in on Ross’s openurl Ruby library discussion. I heard that the basic consensus at the end was that things will stay much as they are now, but there might be a niche for code4lib to provide educational training for libraries. I think this idea came from Dan, and I think it’s a great idea. Hopefully we’ll get a chance to discuss over the coming months.

One of the neatest things I witnessed was Ross Singer spontaneously suggesting a breakout session about designing an openurl library for ruby…and something like 20 people showed up. Not that just any 20 people were there: we had Jeff Young (who wrote OCLCs openurl library), Eric Hellman (who helped write the openurl spec and who just sold his company to OCLC), Todd Holbrook (the software developer behind CUFTS) and Jay (?) one of the software developers beind ExLibris’ SFX product. We had a good discussion, which Ross was able to fascilitate, and I think we came away with some good ideas on how to improve the existing library, and perhaps think about providing a common DOMlike api for openurl implementations.

I could go on and on. Like how great the lightning talks were…for example Terry Reese’s five minute laid back demo of his MARCEdit software that was so polished and amazing I couldn’t believe it. It can query z39.50/SRU targets, and crosswalk to MODs and other metadata formats. Casey Bisson finished the conference on the right note encouraging library software developers to get involved in the technology world outside libraries and to look outwards for cowpaths to pave rather than navel gazing and using only standards developed by libraries. I think he definitely has a point, and that the converse is also true–we should be promoting library standards such as sru/cql in the outside world and encouraging them to pave some of our cowpaths. I was hoping to follow Casey’s talk with my lightning talk about microformats but alas we ran out of time.

All in all I had a great time, and got a chance to meet some really interesting folks (some of whom I got to hang out in Portland with afterwards: Gabriel, Devon, Rob, Aaron). I don’t think it would’ve been possible without the support of people like Art Rhyno, Roy Tennant, Dan Chudnov and of course Jeremy Frumkin who managed to make it just happen. The most important feature of the conference was the size, which was big enough to make it interesting, but small enough to make it easily experienced as a whole, and relaxed enough to be fun. I think that it’s pretty clear that it hit a sweet spot, and that it is highly likely that it will happen again.