@cssquirrel: indeed. On the /> issue, I think it's current status in the spec is about as good as we can expect.