There's no more harm in running OSPF with IPX than RIP/IP with IPX.... The two routing tables are seperate, the two protocols carry different sorts of information, and the two protocols use seperate hello/neighbor/router discovery.
You should be very safe with this.
Deploying OSPF is probably not that hard from RIP. I'd start by putting OSPF on the routers, and setting the admin distance really high on the osp processes. So, everything would be learned through OSPF, but not installed in the routing tables. Next, I'd plan some downtime, if possible, and go through checking each ospf database to make certain all the routing information you're after is there, removing rip, and setting the ospf admin distance back to the default.
To do this, it's best to telnet to the router closest to you, and from there to the next router, and from there to the next router (hop by hop all the way to the opposite adge from your attachment point), and do the changes there, then back out one router, make changes, back out one router, etc. This way you're not relying on routing to get to the routers that need to be changed. That's a bad thing to do when changing your routing protocol. :-)
I've seen multiple networks make this sort of transition. With good planning, and keeping your ear to the ground before and during migration, you should be fine. Note that you have to be careful of router memory if you are going to run two routing processes at the same time, as I suggested above.
Russ.W
