Keeping Partial Duplicates but Ignoring Missings

[hashman]

@mkeintz:

Mark, I may be overlooking something, but to my mind the logic is no different in principle from the following (not using FIND/C function deliberately): 

data _null_ ;                                           
  array rec $1 r1-r8 ("1","2","1","3","4","3","5","2") ;
  length hash $ 8 ;                                     
  do over rec ;                                         
    dup = . ;                                           
    do i = 1 to length (hash) ;                         
      if char (hash, i) = rec then dup = 1 ;            
    end ;                                               
    if not dup then hash = cats (hash, rec) ;           
  end ;                                                 
  put hash= ;                                           
run ;                                                   

Which results in:

hash=12345

The differences are that the real comparison condition is more complex than:

char (hash, i) = rec

and the hash table is used instead of the string HASH. So, basically, we're adding the hash items while eliminating the post-coming dupes in the process, so there's no need to delete anything from the hash when there're no more records.

Kind regards

Paul D.