Question

I need to create a trigger that will add a record to a queue table whenever a record is modified in the main table. The record added to the queue table must contain every field that was modified for that record.

I have this code so far but I don’t think it will work for multiple rows updated:

ALTER TRIGGER [dbo].[tr_EmpHistory]    
ON [dbo].[employeeData]  
FOR UPDATE 
AS 
BEGIN      
DECLARE @FieldsUpdated xml,  
@FieldsUpdated1 varchar(100)       
SELECT @Fieldsupdated1 = ' ' 
SELECT 
@FieldsUpdated1 = @FieldsUpdated1 + ' emp_bankaccountnumber' 
FROM inserted as a,  
deleted as b  
WHERE a.emp_id = b.emp_id             
AND a.emp_bankAccountNumber <> b.emp_bankAccountNumber  
SELECT      
@FieldsUpdated1 = @FieldsUpdated1 + 'emp_salary ' 
FROM inserted as a,        
deleted as b   
WHERE a.emp_id = b.emp_id       
AND a.emp_salary <> b.emp_salary 

SELECT    
@FieldsUpdated1 = @FieldsUpdated1 + 'emp_SSN ' 
FROM inserted as a,  
deleted as b  
WHERE a.emp_id = b.emp_id            
AND a.emp_SSN <> b.emp_SSN  
SELECT 
@FieldsUpdated1 = @FieldsUpdated1 + 'emp_lname ' 
FROM inserted as a,  
deleted as b  
WHERE a.emp_id = b.emp_id           
AND a.emp_lname <> b.emp_lname  
SELECT      
@FieldsUpdated1 = @FieldsUpdated1 + 'emp_fname ' 
FROM inserted as a,  
deleted as b  
WHERE a.emp_id = b.emp_id     
AND a.emp_fname <> b.emp_fname  
SELECT 
@FieldsUpdated1 = @FieldsUpdated1 + 'emp_manager ' 
FROM inserted as a,  
deleted as b  
WHERE a.emp_id = b.emp_id          
AND a.emp_manager <> b.emp_manager  
SELECT @Fieldsupdated =  ( 
SELECT COLUMN_NAME AS Name 
FROM INFORMATION_SCHEMA.COLUMNS  
WHERE TABLE_NAME = 'employeeData' 
AND CHARINDEX(COLUMN_NAME,(ltrim(rtrim(@fieldsupdated1)))) > 0 
FOR XML AUTO, ROOT('Fields') )

INSERT INTO auditEmployeeData( 
audit_emp_id, 
audit_emp_bankAccountNumber,
audit_emp_salary, 
audit_emp_SSN, 
audit_emp_lname, 
audit_emp_fname, 
audit_emp_manager, 
ColumnsUpdated ) 

SELECT emp_id, 
emp_bankAccountNumber, 
emp_salary,    
emp_SSN,   
emp_lname,  
emp_fname,     
emp_manager,       
@FieldsUpdated   
FROM INSERTED  
END
GO 

If I am understanding this correctly, if one record had last name updated and another record got the first name updated at the same time then both records would be logged as having both first and last named changed. Is that correct? If so, how could I make it work correctly without using a cursor?

The only way I can think to solve the problem is to use a cursor but I know that is not a good idea. Any help would be appreciated.

Thanks!

Answer 1

You can use CASE to build a value row-by-row showing the changed columns:

select i.emp_id,
  case when i.foo <> d.foo then ',foo' else '' end +
  case when i.bar <> d.bar then ',bar' else '' end as changedcolumns
  from inserted as i inner join
    deleted as d on d.emp_id = i.emp_id

Some additional fiddling about can eliminate the extra delimiter.