I saw version 1.92 of this on CPAN, and it immediately struck me that I wished there was a unique() and a nonunique() function in the module. The function unique() would simply return the next tuple that did not have any elements in common, and nonunique() would return the next tuple that did have an element in common. Sets (1, 2, 3) and (2,3,4) would return seven unique tuples and two nonunique tuples. Even better would be implementing this as a global flag also, so that it affected all other functions, such as next(), previous() and random(), by passing it as part of the use statement.