RETAIN statement anomaly on Hadoop Data

[Tom]

Is the problem reproducible? 

If so then just ask your data step to show you the records where you are seeing the skip.

Note: Do NOT overwrite your input as it will make debugging very hard.

data WANT;
  set have;
  retain t 0;
  select;
    when (amount NE 0) t=t+1;
    otherwise;
  end;
  if t in (3615 3616 119401 119402) then put _all_ ;
run;