Things learned today:

• Although the db structure is quite confusing (lots of numeric ids, data spread across a lot of tables), it usually turns out that it's easiest to do data management with SQL; only single records can be manipulated easily in the web interface. I've been working with the over 500 records for Nanaimo, and my process has been:
  1. Dump the db as XML.
  2. Change one record in the way I want to change it.
  3. Dump the db XML again.
  4. Diff the two XMLs.
  5. Write SQL to accomplish the same change in all the other records.
  After this kind of manipulation, though, the web interface will not show the results until HR in Sysadmin runs

  $ php symfony search:populate 

  which refreshes the indexes.
• All current records are in draft, so you can't see them unless you're logged in; I've created a "tester" user, and the other project folks are using that to view the data. No feedback as yet.
• I've worked on the Cumberland data, which came in as Access dbs; exported it to Excel, then generated TSV from that with LO Calc. I have a much better XSLT for this one, learning from the last one, and it seems likely that we could create a nice generic one that could be customized through an external file.