Question

I have a Personal table with a LastName column and a MaybeUniqueID.
I want to put in output a table with a LastName column where the counter set on to the column MaybeUniqueID gives more than 1 row.

• I would like to do everything in one unique run, avoiding mid-step outputs.
• I prefere not using temporary table or table variables, otherwise I would like to use at most one table variable (not temporary tables), but I think this should not be necessary.
• I am using Microsoft SQL Server 2005.

I tried different scenarios with different SQL statements like HAVING or GROUP BY, but I failed to get the outcome I am looking for.
Please have a look at the following not-working summary test:

SELECT LastName
FROM   Personal
       JOIN
              (SELECT  MaybeUniqueID AS ID2,
                       COUNT(*)      AS CNT
              FROM     Personal
                       --WHERE CNT > 1
              GROUP BY MaybeUniqueID
              HAVING   cnt > 1
              ) AS MultiMaybeUniqueID
       ON     Personal.MaybeUniqueID = MultiMaybeUniqueID.ID2

Answer 1

HAVING cnt > 1 should be HAVING COUNT(*) > 1.

Column aliases can only be referenced in the ORDER BY clause not the HAVING clause.

Though you could also use

;WITH T AS
(
SELECT LastName,
       COUNT(*) OVER (PARTITION BY MaybeUniqueID) AS Cnt
FROM Personal
)
SELECT LastName
FROM T 
WHERE Cnt > 1