Actually, Oracle is clever enough to maintain currval for your existing session so you will get the correct value even if someone else is accessing or updating the sequence.

Thanks.
Krishantha.

Here is my trigger.

CREATE TABLE test1 (id NUMBER PRIMARY KEY,name VARCHAR2(30) NOT NULL)
/
CREATE SEQUENCE test1_sequence START WITH 1 INCREMENT BY 1
/
CREATE OR REPLACE TRIGGER test1_trigger
AFTER INSERT
ON test1
REFERENCING NEW AS NEW
FOR EACH ROW
BEGIN
SELECT test1_sequence.nextval INTO :NEW.ID FROM dual;
END;
/
commit;
/

After trying to insert data to the table with NULL name value. ID sequence got changed.

SQL> select * from test1;

ID NAME
———- ——————————
1 kkTEST
2 kkTEST11
3 kkTEST12
5 kkTEST14555

I update the “before insert” as “after insert” and executed the trigger. This time I got the following error.

ORA-04084: cannot change NEW values for this trigger type

Any help will be greatly appreciated.

Thanks,
Krishantha.

How are you activating the trigger? You might be able to prevent false incrementing by changing to an ‘after insert’ trigger. More likely you’ll want to make the trigger check the name column and raise an error if it’s null.

Hope this helps.

Thanks for your valuable tips. My table A has two columns which are ID and Name. I used sequence to increment the ID number sequentially. “Name” column is with not null constraint. My problem in here is that if a user insert data into table “A” without giving a value for “Name” column then the trigger will fire and sequence value will be increment by 1. in the next time if I insert name value,then ID numbers are not sequential. How I can handle this problem from DB level.

Thanks,
Krishantha.

The only way I know to do this is to create a sequence for each table. There are few instances where you would want to use one sequence for two different tables. Sequences take up very little resources so there is no reason not to have one per table.

Now the id for T1 will be 1 and id for T2 will start from 2

ofcourse I can avoid this situation by having 2 sequencer one each for T1 and T2, but just wanted to check if there is a alternative

using this way:

insert into mytable (a, b) values (1, (select NVL(max(b),0)+1 from mytable where a=1));

will ensure it works for the first time and everytime
as Jon rightly said, this query will kill the performance.