Began work on some code to detect and evaluate matches between records in the original CGWP data and the LAC records, using XSLT. The logic is working well: we have a score between 0 and 15 for any comparison between records, based on ids, names, birth dates, regimental numbers and so on, and we're able to collect and report the best matches for any pair of records. We can run through the CGWP data and report these matches in a processable form. We hit a Saxon bug that was puzzling (it worked in Oxygen but failed at the command line with Saxon; this error went away after upgrading Saxon to this month's release). However, when we attempt to process large numbers of records, we hit a bug with one specific record which appears rather mysterious; we'll come back and look at this in more detail tomorrow. It seems likely that processing the whole set will take three days, so we need to have a rock-solid process before we start, and we need to be able to restart in the middle after a problem happens.