Charles was able to determine that the memory leak was actually a JDK 1.4.1 StringBuffer bug. We seem to have fixed it by upgrading to JDK 1.4.3. Hooray!

I’m now working on a statistics module so we can see how people use the site. Nothing fancy… just dump results from the database into Excel spreadsheets so that people can manipulate the data on their own. This should be a useful addition to the data we can get from WebTrends.

In particular, my module should make it easier to parse out which licensed resources are being used with which search terms, which might prove an interesting area for future work/improvements. Right now we are using data from our database, but I wonder how much can be manipulated out of the WebTrends data… perhaps with a little post-processing to make it more user-friendly.

Everything else with the federated search seems pretty stable at this point.