I think Howard Bullock has a pretty good solution with his HASH (UDF) function but I still didn't manage to get the script to work with HASH (UDF). Perhaps the complexity is too high...