We should be trying OPENSTEP in parallel, with NeXTSTEP as the "upstream" with prime focus. Most of the time I think they will both behave the same.