Question

I’ve read some questions/answers here on Stack but couldn’t find anything like what I need.
Let’s say I have the following table:

col0  col1  col2  col3  col4
----  ----  ----  ----  ----
8     1     a     b     c
8     2     a     b     c
8     3     a     b     c
9     1     a     b     c
9     2     a     b     c

And my software does:

INSERT INTO testtable ([col0],[col1],[col2],[col3],[col4]) 
  VALUES ('8','4','a','b','c')

How can I create a trigger like this pseudo-code:

On insert when col1 = '4'
  delete existing rows where col0 is the same (8 in this case)
   ** except for the new row I've just added **

Answer 1

It sounds more appropriate to perform this in an instead of trigger. If the goal is what I assume: when a new row is inserted with col0 = 8, you want to delete all other rows with that same key, yes?

CREATE TRIGGER dbo.testtable_instead_insert
ON dbo.testtable
INSTEAD OF INSERT
AS
BEGIN
  SET NOCOUNT ON;

  DELETE t 
    FROM dbo.testtable AS t
    INNER JOIN inserted AS i
    ON t.col0 = i.col0
    WHERE i.col1 = '4';

  INSERT dbo.testtable(col0,col1,col2,col3,col4)
    SELECT col0,col1,col2,col3,col4
      FROM inserted;
END
GO

The question is what rule do you want to follow if an insert attempts to add multiple rows with the same col0 value? Imagine the insert statement is:

INSERT dbo.testtable ([col0],[col1],[col2],[col3],[col4]) 
SELECT '8','4','a','b','c'
UNION ALL 
SELECT '8','3','c','d','e';