We'll probably keep the SDM as a 3.x instance at first, now that it appears that the major performance problems with OpenACS 4/PG have been solved (which means we can start thinking seriously again about the move).
This will be ugly but I don't see the SDM for 4.x getting done in time.
