Question

That is my table #EmployeeTemp.

What I would like to do is retrieving data like that below.

So first of all, I use below query to retrieve data as I want.

CREATE TABLE #EmployeeTemp
(Employee_Name VARCHAR(50),
Confirmed_Date DATETIME,
Field_Type VARCHAR(50),
Field_Value VARCHAR(50))

INSERT INTO #EmployeeTemp VALUES ('John', '2009/1/1', 'Title', 'Programmer');
INSERT INTO #EmployeeTemp VALUES ('John', '2009/1/1', 'Department', 'IT Department');
INSERT INTO #EmployeeTemp VALUES ('John', '2011/1/1', 'Title', 'Sr. Programmer');
INSERT INTO #EmployeeTemp VALUES ('David', '2010/1/1', 'Title', 'Software Engineer');
INSERT INTO #EmployeeTemp VALUES ('David', '2010/1/1', 'Department', 'IT Department');

SELECT * FROM #EmployeeTemp

SELECT 
MyEmployee.Confirmed_Date,
MyEmployee.Employee_Name,   
Title = MAX(MyEmployee.Field_Value),
Department = (SELECT 
                    Field_Value
                FROM #EmployeeTemp
                WHERE Field_Type = 'Department'
                AND Employee_Name = MyEmployee.Employee_Name
                GROUP BY Employee_Name, Field_Value)
FROM #EmployeeTemp AS MyEmployee
WHERE MyEmployee.Confirmed_Date IN (SELECT MAX(Confirmed_Date) FROM #EmployeeTemp GROUP BY Employee_Name)
GROUP BY MyEmployee.Employee_Name, MyEmployee.Confirmed_Date
ORDER BY MyEmployee.Confirmed_Date DESC

It worked as I wanted. But what I would like to ask is could you please give me any suggestion to write query more professional way.

Every suggestion will be appreciated.

I am using Microsoft SQL Server 2008 version.

Answer 1

This question is very subjective, so what in my opinion is a more elegant solution, may be a less elegant solution in someone else’s opinion. However, this is the approach I would adopt, using the SQL-Server PIVOT Functionality.

;WITH Employees AS
(   SELECT  *, 
            MAX(Confirmed_Date) OVER(PARTITION BY Employee_Name, Field_Type) [MaxFieldConfirmedDate],
            MAX(Confirmed_Date) OVER(PARTITION BY Employee_Name) [MaxConfirmedDate]
    FROM    #EmployeeTemp
)
SELECT  *
FROM    (   SELECT  MaxConfirmedDate [Confirmed_Date], Employee_name, Field_Type, Field_Value
            FROM    Employees
            WHERE   Confirmed_Date = MaxFieldConfirmedDate
        ) data
        PIVOT
        (   MAX(Field_Value)
            FOR Field_Type IN ([Title], [Department])
        ) Pvt

The reason I prefer this approach is firstly it is less work on the server (when checking actual execution plans for your query compared to this one, this one had a cost of 27% relative to the batch, as opposed to the 73% cost of yours). Secondly, if you need to add more fields, it is simply a case of extending the following line:

FOR Field_Type IN ([Title], [Department])

e.g. with the following extra data:

INSERT INTO #EmployeeTemp VALUES ('John', '2011/1/1', 'Test', 'Test');
INSERT INTO #EmployeeTemp VALUES ('David', '2010/1/1', 'Test', 'Test');

you would only need to make the line:

FOR Field_Type IN ([Title], [Department], [Test])

to get the extra column required in the output.