Note:- When ever, we excute any DML (INSERT or UPDATE or DELETE) Operations in the PL/SQL block then It will move from PL/SQL Block to SQL Engine to excute the DML Operation in SQL Engine. And if we have some SELECT statement in the PL/SQL Block, then it will move from SQL Enginer to get the data from SQL Enginer to the PL/SQL Block. For the SELECT statement, we use the BULK COLLECT to record the Iteration from the SQL Enginer to the PL/SQL Block.
If I have to put my self in other word then, I should say, BULK COLLECT will process group of SELECT statements at one short.
Similarly for the DML Operations, we use FORALL Concept, please refer the FORALL Concept for more details.
Note:- This Example is based on the EMP table in the SCOTT schema. Here say, EMP table have 90 Records and if we do it in the Normal way (Without BULK COLLECT concept, then It will iterate SQL enginer to the PL/SQL Block for 90 Times). By using BULK COLLECT, we can completly reduce the Iterations.
In the following Example, I have limited to 10, so for 90 records, this will iterate for 9 times.
type array is table of number index by binary_integer;
cursor c is select empno from emp;
fetch c bulk collect into l_data limit 10;
if ( c%notfound )
( ‘Cursor returned NOT FOUND but array has ‘ || l_data.count
|| ‘ left to process’ );
( ‘We have ‘ || l_data.count
|| ‘ to process’ );
exit when c%notfound;
Returning clause with the Bulk collect with the DML opeartions:-
DELETE FROM emp WHERE num = 30
RETURNING empno, ename BULK COLLECT INTO p_num, p_ename;