Working with SA, did some test encoding of a range of different directory entries, attempting to nail down our preferred encoding strategies ahead of developing the ODD file. We have dealt with most of the phenomena we expect to encounter. There's going to be a lot of heavy Schematron and content constraint going on to keep this nailed down, I think. It's an interesting mix between representation of the primary source (so we don't miss something that might later prove to be important) and database-style tagging (so everything taggable is tagged predictably and precisely)