What is the difference between the two sql statements?

Question

Good Day

All we are trying to do is inside a trigger make sure the user is not inserting two fees that have 'alone' in the name. Those fees need to be handled individually.

For some reason, it appears the top section of sql quit working two weeks ago. To get around it I recoded it the second way and get the correct results. What I am confused is why would the first portion seemed to have worked for the last several years and now it does not?

SELECT  @AloneRecordCount = count(*) 

FROM  inserted i 
      INNER JOIN deleted d on i.id = d.id 

WHERE  i.StatusID = 32 
      AND d.StatusID <> 32 
      AND i.id IN  
      (SELECT settlementid FROM vwFundingDisbursement fd
      WHERE fd.DisbTypeName LIKE  '%Alone'
      AND fd.PaymentMethodID = 0)


SELECT  @AloneRecordCount = count(i.id) 

FROM    inserted i INNER JOIN
        deleted d on i.id = d.id
        JOIN vwFundingDisbursement fd on i.id = fd.settlementid

WHERE   i.StatusID = 32 
        AND d.StatusID <> 32  
        AND fd.DisbTypeName like '%Alone'
        AND fd.PaymentMethodID = 0

this is on SQL Server 2005
there is no error, instead the top statement will only return 1 or zero
while the bottom statement will return the actual number found.

Answer 1

Have you got a null vwFundingDisbursement.settlementid value? Has that appeared recently?

Answer 2

count(*)
count(i.id)

will return different counts based on NULL values

http://stackoverflow.com/questions/59322/in-sql-whats-the-difference-between-count-and-countx

Answer 3

A schema (or at least how the view is created) would help, but here's a guess...

If you are looking for multiple rows in vwFundingDisbursement with the value "Alone" in the distribution type name, then a JOIN is going to return multiple rows as your source table (INSERTED) joins to multiple rows in the view. If you use IN though, SQL doesn't care if it returns multiple matches, it's only going to give you one row.

As an example:

CREATE TABLE dbo.Test_In_vs_Join1
(
     my_id     INT     NOT NULL
)

CREATE TABLE dbo.Test_In_vs_Join2
(
     my_id     INT     NOT NULL
)

INSERT INTO dbo.Test_In_vs_Join1 (my_id) VALUES (1)
INSERT INTO dbo.Test_In_vs_Join1 (my_id) VALUES (2)
INSERT INTO dbo.Test_In_vs_Join1 (my_id) VALUES (3)
INSERT INTO dbo.Test_In_vs_Join1 (my_id) VALUES (4)
INSERT INTO dbo.Test_In_vs_Join1 (my_id) VALUES (5)

INSERT INTO dbo.Test_In_vs_Join2 (my_id) VALUES (1)
INSERT INTO dbo.Test_In_vs_Join2 (my_id) VALUES (1)
INSERT INTO dbo.Test_In_vs_Join2 (my_id) VALUES (2)
INSERT INTO dbo.Test_In_vs_Join2 (my_id) VALUES (3)
INSERT INTO dbo.Test_In_vs_Join2 (my_id) VALUES (3)

SELECT
     T1.my_id,
     COUNT(*)
FROM
     dbo.Test_In_vs_Join1 T1
INNER JOIN dbo.Test_In_vs_Join2 T2 ON
     T2.my_id = T1.my_id
GROUP BY
    T1.my_id

SELECT
     T1.my_id,
     COUNT(*)
FROM
     dbo.Test_In_vs_Join1 T1
WHERE
    T1.my_id IN (SELECT T2.my_id FROM dbo.Test_In_vs_Join2 T2)
GROUP BY
    T1.my_id

On a side note, burying a column inside of another column like this is a violation of the normalized form and just asking for problems. Performing this kind of business logic in a trigger is also a dangerous path to go down as you're finding out.

Answer 4

I assume this MSSQL(?) and it is handling an UPDATE statement since you're joining both the inserted/deleted tables?

Which table is this trigger on?