Didn't we decide to go with binary search for big integer case statements? The data from the GNU Eiffel paper seemed pretty convincing that it is often much getter and basically never much worse. I was confused initially by this message trying to see what the importance of IntInf.int switches was (i.e., BIG integers).