There's no documentation currently on what formats we support and how we split things up for parallel processing. Related: #337