Can you explain more what you are trying to do?
If you want to "join" the dataset on exact matches and at least one of the datasets does not have duplicate keys then it is going to be much easier to use a normal data step MERGE instead of trying to debug what SQL is doing. You can use the IN= dataset option to if you want to ignore observations that are not in one of the datasets.
data want;
merge left(in=in1) right(in=in2);
by key;
if in1;
run;
... View more
