Spent the day working on getting the code to validate in the locations files. Found lots of problems caused by my conversion, which were easy to fix, and some other things which we'll have to watch out for elsewhere in the database, such as date/@value instead of date/@when, and person[@type='person'] instead of name[@type='person']. There are also two files (TRAN1 and TRAN2) which represented in the locations files (because they were in the index), but which actually have writeups in the topics subcollection; waiting for JJ's approval to move those writeups into the locations files.

GN and I also did some analysis of the "original" map setup. The XSLT output just creates an HTML map element, and the rest is done by JavaScript, so it should be fairly straightforward to reproduce it.