If i try left outer or full outer joins i just end up with duplicates etc. I am not sure if the output i am after is possible (my SQL skills aren't amazing). I did think of trying to add uniqueness to each and row and match on that but not sure how i would automate adding a count to each group of duplicates to do this either.
how are you generating these datasets? if it is generated using SQL we can try to generate the unique ids using sql and even generate the final joined data in sql. post the existing sql.