I'm a bit confused here.

Is the proposal that the library design exposes to the programmer the following type synonyms:

type IO a = ST IORegion a type IORef a = STRef IORegion a

I think it is sound to do so. The advantage would be that newIORef and newSTRef would be the same (newRef, perhaps); ditto readRef, writeRef etc.

And possibly type IOArray ix e = STArray IORegion ix e type IOUArray ix e = STUArray IORegion ix e and lots of duplication in the overloaded array support goes away. I'd be very careful that we're not burning any bridges before doing this, though. Cheers, Simon