Pl sql cursor

Question

hi to everyone
          this is my coding and i want you to make a slight adjustment to print in a format which i required..
DECLARE
department_no   emp.empno%TYPE;
   CURSOR cur
   IS
      SELECT *
        FROM emp
  order by deptno;
      BEGIN
   FOR i IN cur
   LOOP
      department_no := i.deptno;
      sys.DBMS_OUTPUT.put_line ('department  no is' || department_no);
      sys.DBMS_OUTPUT.put_line ('employee first name is' || i.ename);
      sys.DBMS_OUTPUT.put_line ('-------------------------');
   END LOOP;
   END;
the format i required is:
department  no is10
employee first name isCLARK
employee first name isKING
employee first name isMILLER
department  no is20
employee first name isJONES
employee first name isFORD
employee first name isSMITH
employee first name isSCOTT
department  no is30
employee first name isWARD
employee first name isTURNER
thanks..

Answer 1

Hi,
What's wrong with your current code?  The only thing wrong is that it displays the department before every row (not just the first row in each department) and the line of hyphens after every row (not just the last line in each department).  Use IF statements to display these thing only when a department changes.
How can you know when a department changes?  You could add analytic functions to the query.  ROW_NUMBER can assign  numbers 1, 2, 3, ... to the rows in each department, either in ascending or descending order.  The row numbered 1 in ascending order is the first row in its department, and the row numbered 1 in descending order is the last row.
Instead of using analytic functions, you can create a local variable the your code, that remembers the last department.  If the current depatment is deifferent from the last department, then you need a line of hyphens, and a new department header.  You'' always need an additional department header before the very first row, and a row of hyphens after the very last row, too.
Either way, give it a try.  If you get stuck, post your revised code.
You could get the results you want using GROUP BY GROUPING SETS, without using PL/SQL at all, but it's more complicated and harder to maintain.