Tuesday, March 4, 2008

An Aquifer Perspective on Code4Lib Conf 2008

It's a small well run tightly focused forum for programming efforts in support of libraries. With a single track, primary talks limited to 20 minutes talks, then ending each day with an hour of 5 minute lightning talks you can't help feeling in touch with what's going on in the library programming community. See http://code4lib.org/conference/2008/
The trend that stood out most for me is the number of projects implementing catalogs in one form or another. The software tools necessary to create a catalog are more robust and easier to use, borne out by our own project. Very little of the presentation time on these things was spent bemoaning the difficulties of master the different metadata formats, the problems, whatever they are, seem secondary to the issues of getting real time information on holdings, circulation status, etc. My sense is that the continuing movement towards keyword and faceted searching via SOLR/Lucene and like engines is sufficient in the minds of this group (at least) to address concerns about how the quality and differences of the underlying metadata affect the ability to search and present it.
Not that metadata issues are dead, two rather pointed talks on the on-going RDA effort show that to be far from the case. My naïve view is that all metadata efforts would be immensely helped by employing programmers on the metadata team up front and charging them with delivering, upon release of the specification, a reference implementation, any reference implementation. Secondly, (and I can't quite believe I am saying this) more exuberant use of the specificity that xml can provide. As an example, Karen Coyle, one of the keynote speakers, described the inability to achieve closure on the issue of encoding the author and publisher when what appears on cover page is wrong. Since this seems to be important, I say, go for a little tag richness and provide an encoding for both pieces. What will make this work is that the programmer working on the reference implementation will need to know the preferred order for presenting this detail via DC or RSS which is all that most of the world will ever see of it. It's the right solution, the information is not lost but implementers will have a clue what the specification architects had in mind.
In short: great conference, lovely city, nice weather, a beer town of mythic proportions.

No comments: