A variation of this question popped up twice in the SQL programming newsgroup since yesterday, this means it is time for a blogpost.
Let's say you have a proc like this
CREATE PROC prUpdateTable
@Salesman int = -1
AS
..........
If the user calls the proc like this exec prUpdateTable null then @Salesman will be null, if the user calls the proc like this exec prUpdateTable then the value of @Salesman will be -1. In both of this cases you don't want to change the value of the column. What can you do?
You can use a combination of NULLIF and COALESCE to handle this. Your update statement would look like this
UPDATE table
SET Column = COALESCE(NULLIF(@variable,-1),Column)
Here is some code to demonstrate that
CREATE TABLE #foo (id int,salesman int)
INSERT #foo VALUES(1,1)
INSERT #foo VALUES(2,1)
INSERT #foo VALUES(3,1)
INSERT #foo VALUES(4,1)
SELECT * FROM #foo
-------------
1 1
2 1
3 1
4 1
DECLARE @salesman int
SELECT @salesman = 5
--Column value will change to 5
UPDATE #foo
SET salesman = COALESCE(NULLIF(@salesman,-1),salesman)
WHERE ID =1
--Column value won't change
SELECT @salesman = -1
UPDATE #foo
SET salesman = COALESCE(NULLIF(@salesman,-1),salesman)
WHERE ID =2
--Column value won't change
SELECT @salesman = NULL
UPDATE #foo
SET salesman = COALESCE(NULLIF(@salesman,-1),salesman)
WHERE ID =3
--Column value will change to 3
SELECT @salesman = 3
UPDATE #foo
SET salesman = COALESCE(NULLIF(@salesman,-1),salesman)
WHERE ID =4
--And here is the output, as you can see when @salesman was -1 or NULL the table did not get updated
SELECT * FROM #foo
-------------
1 5
2 1
3 1
4 3
DROP TABLE #foo
As you can see only the first and the last update statement changed the value of the salesman column
Of course you would never do this if you were to update only one column, you would skip the update instead. If you have to update multiple columns then this is something you can use instead of writing a bunch of dynamic SQL or nested IF statements.