The sorting literature pays attention to stability, and some practical applications need stability, so it's good that the SRFI provides guaranteed-stable sort procedures.
The same is true of sort algorithms with deterministic (i.e. non-randomized) run times. So, could we add procedures that guarantee a deterministic worst case time bound?