Oracle SQL触发更新问题-研究/学习

Question

I have created a copy of an existing database (I_Customer) with the below code:

CREATE TABLE N_CUSTOMER AS SELECT * FROM I_CUSTOMER;

I have created the below table that will act as a log for modifications to the N_Customer table:

CREATE TABLE CUSTOMER_CHANGELOG 
( cust_no NUMBER(5),
cust_name VARCHAR2(20),
contact VARCHAR2(20),
log_date DATE);

I have also created (with help from a stack member) the below trigger that will fire after the N_Customer table is edited and write some specific fields (new and old) tot he Customer_Changelog table:

CREATE OR REPLACE 
TRIGGER customer_up_tr
AFTER UPDATE ON n_customer
FOR EACH ROW
WHEN (OLD.contact <> 1 AND NEW.contact = 1 OR OLD.cust_name <> 1 AND NEW.cust_name = 1)
BEGIN
      INSERT INTO customer_changelog (cust_no, cust_name, contact, log_date) VALUES (:OLD.cust_no, :OLD.cust_name, :OLD.contact, sysdate);
      INSERT INTO customer_changelog (cust_no, cust_name, contact, log_date) VALUES (:NEW.cust_no, :NEW.cust_name, :NEW.contact, sysdate);
END;

Now when I use the below code to edit the N_Customer table:

UPDATE N_CUSTOMER
SET cust_name = 'Peter Davis', contact = 'Sam Bogdanovich'
WHERE cust_no = 2338;

I get this error:

UPDATE N_CUSTOMER
*
ERROR at line 1: 
ORA-01722: invalid number 

Now my data types etc all match so I am not sure what could be causing this.

Any ideas would be much appreciated.

Answer 1

The problem is in this line of the trigger:

(OLD.contact <> 1 AND NEW.contact = 1 OR OLD.cust_name <> 1 AND NEW.cust_name = 1)

Contact and cust_name fields are VARCHAR2, but the trigger is comparing them to a number.

This simple example will fail with the same error:

SELECT 1 FROM DUAL
WHERE 'abc' <> 1;

Answer 2

The expression OLD.contact <> 1 is comparing the character column contact to a number (This is also known as "comparing apples to oranges").

Because the conversion of a number to a character is ambigous (it could be converted to '01', '001' ...) Oracle will do it the other way round and convert the character value to a number (because that is non-amgigous '01' , '001' will both result in the number 1`).

You apparently have non-numeric values in the contact column (which is to be expected based on the column's name and it's data type) the conversion fails.

The simple rule is:

Never rely on implicit type casting

Use character literals for characters and number literals for numbers. 1 is a number. '1' is a character literal.

Having said all that, you need to change this line:

OLD.contact <> 1 AND NEW.contact = 1 OR OLD.cust_name <> 1 AND NEW.cust_name = 1

to

OLD.contact <> '1' AND NEW.contact = '1' OR OLD.cust_name <> '1' AND NEW.cust_name = '1'