Question on SQL Grouping

Question

Hi Team,

I am trying to achieve the following without using sub query.

For a funding, I would like to select the latest Letter created date and the ‘earliest worklist created since letter created’ date for a funding.

FundingId Leter (1, 1/1/2009 )(1, 5/5/2009) (1, 8/8/2009) (2, 3/3/2009) 

FundingId WorkList (1, 5/5/2009 ) (1, 9/9/2009) (1, 10/10/2009) (2, 2/2/2009) 

Expected Result -

FundingId Leter WorkList (1, 8/8/2009, 9/9/2009)

I wrote a query as follows. It has a bug. It will omit those FundingId for which the minimum WorkList date is less than latest Letter date (even though it has another worklist with greater than letter created date).

CREATE TABLE #Funding(
[Funding_ID] [int] IDENTITY(1,1) NOT NULL,
[Funding_No] [int] NOT NULL,
CONSTRAINT [PK_Center_Center_ID] PRIMARY KEY NONCLUSTERED ([Funding_ID] ASC)
) ON [PRIMARY]

CREATE TABLE #Letter(
[Letter_ID] [int] IDENTITY(1,1) NOT NULL,
[Funding_ID] [int] NOT NULL,
[CreatedDt] [SMALLDATETIME],
CONSTRAINT [PK_Letter_Letter_ID] PRIMARY KEY NONCLUSTERED ([Letter_ID] ASC)
) ON [PRIMARY]

CREATE TABLE #WorkList(
[WorkList_ID] [int] IDENTITY(1,1) NOT NULL,
[Funding_ID] [int] NOT NULL,
[CreatedDt] [SMALLDATETIME],
CONSTRAINT [PK_WorkList_WorkList_ID] PRIMARY KEY NONCLUSTERED ([WorkList_ID] ASC)
) ON [PRIMARY]

SELECT F.Funding_ID,
Funding_No, 
MAX (L.CreatedDt),
MIN(W.CreatedDt)
FROM #Funding F
INNER JOIN #Letter L ON L.Funding_ID = F.Funding_ID
LEFT OUTER JOIN #WorkList W ON W.Funding_ID = F.Funding_ID
GROUP BY F.Funding_ID,Funding_No
HAVING MIN(W.CreatedDt) > MAX (L.CreatedDt)

How can I write a correct query without using subquery?

Please help

Thanks

Lijo

Answer 1

Your question is How can I write a correct query without using subquery?

But you're not using a sub-query... so you already have your answer.

Answer 2

using derived tables is as good as it gets:

OP's tables:

CREATE TABLE #Funding(
[Funding_ID] [int] IDENTITY(1,1) NOT NULL,
[Funding_No] [int] NOT NULL,
CONSTRAINT [PK_Center_Center_ID] PRIMARY KEY NONCLUSTERED ([Funding_ID] ASC)
) ON [PRIMARY]

CREATE TABLE #Letter(
[Letter_ID] [int] IDENTITY(1,1) NOT NULL,
[Funding_ID] [int] NOT NULL,
[CreatedDt] [SMALLDATETIME],
CONSTRAINT [PK_Letter_Letter_ID] PRIMARY KEY NONCLUSTERED ([Letter_ID] ASC)
) ON [PRIMARY]

CREATE TABLE #WorkList(
[WorkList_ID] [int] IDENTITY(1,1) NOT NULL,
[Funding_ID] [int] NOT NULL,
[CreatedDt] [SMALLDATETIME],
CONSTRAINT [PK_WorkList_WorkList_ID] PRIMARY KEY NONCLUSTERED ([WorkList_ID] ASC)
) ON [PRIMARY]

OP's sample data:

INSERT INTO #Funding (Funding_No) VALUES (1)
INSERT INTO #Funding (Funding_No) VALUES (2)

INSERT INTO #Letter (Funding_ID,CreatedDt) VALUES (1,'1/1/2009')
INSERT INTO #Letter (Funding_ID,CreatedDt) VALUES (1,'5/5/2009')
INSERT INTO #Letter (Funding_ID,CreatedDt) VALUES (1,'8/8/2009')
INSERT INTO #Letter (Funding_ID,CreatedDt) VALUES (2,'3/3/2009')

INSERT INTO #WorkList (Funding_ID,CreatedDt) VALUES (1, '5/5/2009')
INSERT INTO #WorkList (Funding_ID,CreatedDt) VALUES (1, '9/9/2009')
INSERT INTO #WorkList (Funding_ID,CreatedDt) VALUES (1, '10/10/2009')
INSERT INTO #WorkList (Funding_ID,CreatedDt) VALUES (2, '2/2/2009')

The table CREATEs look like TSQL, but no version was given, so a CTE could have been used as well. However, this uses derived tables:

SELECT
    dt.Funding_ID,LCreatedDt,MIN(CreatedDt) AS WCreatedDt
    FROM (SELECT
              f.Funding_Id,l.LCreatedDt
              FROM #Funding    f
              LEFT OUTER JOIN (SELECT
                                   Funding_ID,MAX(CreatedDt) AS LCreatedDt
                                   FROM #Letter
                                   GROUP BY Funding_ID
                              ) l ON f.Funding_ID=l.Funding_ID
         ) dt
    LEFT OUTER JOIN #WorkList w ON dt.Funding_ID=w.Funding_ID
    WHERE w.CreatedDt>dt.LCreatedDt
    GROUP BY dt.Funding_ID,LCreatedDt

OUTPUT:

Funding_ID  LCreatedDt              WCreatedDt
----------- ----------------------- -----------------------
1           2009-08-08 00:00:00     2009-09-09 00:00:00

(1 row(s) affected)

to preempt any people claiming that my query uses a subquery, read this article first on Subquery Fundamentals: http://msdn.microsoft.com/en-us/library/aa213252(SQL.80).aspx

A subquery is a SELECT query that returns a single value and is nested inside a SELECT, INSERT, UPDATE, or DELETE statement, or inside another subquery. A subquery can be used anywhere an expression is allowed.