+3  A: 

I personally hate pivots- hard to read and unweidly.

CREATE TABLE #test
(
    WeekNo int,
    [DayOfWeek] int,
    FromTime time,
    ToTime time
    )

INSERT INTO #test
SELECT 1,2,'10:00','14:00'
UNION ALL
SELECT 1,3,'10:00','14:00'
UNION ALL
SELECT 2,3,'08:00','13:00'
UNION ALL
SELECT 2,4,'09:00','13:00'
UNION ALL
SELECT 2,5,'14:00','22:00'
UNION ALL
SELECT 3,1,'06:00','13:00'
UNION ALL
SELECT 3,4,'06:00','13:00'
UNION ALL
SELECT 3,5,'14:00','22:00'

SELECT WeekNo, 
    MAX(CASE WHEN DayOfWeek = 1 THEN FromTime ELSE NULL END)  AS Start1,
    MAX(CASE WHEN DayOfWeek = 1 THEN ToTime ELSE NULL END)  AS End1,
    MAX(CASE WHEN DayOfWeek = 2 THEN FromTime ELSE NULL END)  AS Start2,
    MAX(CASE WHEN DayOfWeek = 2 THEN ToTime ELSE NULL END)  AS End2,
    MAX(CASE WHEN DayOfWeek = 3 THEN FromTime ELSE NULL END)  AS Start3,
    MAX(CASE WHEN DayOfWeek = 3 THEN ToTime ELSE NULL END)  AS End3,
    MAX(CASE WHEN DayOfWeek = 4 THEN FromTime ELSE NULL END)  AS Start4,
    MAX(CASE WHEN DayOfWeek = 4 THEN ToTime ELSE NULL END)  AS End4,
    MAX(CASE WHEN DayOfWeek = 5 THEN FromTime ELSE NULL END)  AS Start5,
    MAX(CASE WHEN DayOfWeek = 5 THEN ToTime ELSE NULL END)  AS End5,
    MAX(CASE WHEN DayOfWeek = 6 THEN FromTime ELSE NULL END)  AS Start6,
    MAX(CASE WHEN DayOfWeek = 6 THEN ToTime ELSE NULL END)  AS End6,
    MAX(CASE WHEN DayOfWeek = 7 THEN FromTime ELSE NULL END)  AS Start7,
    MAX(CASE WHEN DayOfWeek = 7 THEN ToTime ELSE NULL END)  AS End7
    FROM #test
    GROUP BY WeekNo

And it'll blow the socks off of a pivot; performance wise.

Mike M.
Looks like a draw between the two approaches performance wise.
Martin Smith
@Martin Smith - Mine has to UNPIVOT first. Because of that, I would have thought the traditional "pivot" would have been faster.
Cade Roux
Can't really tell based on this data set; would need to create some large test data. Even with this tiny data the cross tab is performing better (0ms vs 6ms). But check out the execution plans if you're interested in the differences.
Mike M.
I did already. Both showing 50% cost when run together. Both 1 table scan and remarkably similar execution plans. A few extra operators in the PIVOT/UNPIVOT but not contributing anything significant to the cost. Socks mildly ruffled perhaps but decidedly not blown off!
Martin Smith
I'm not able to access SQL right now but i'm fairly certain with a couple million rows we can see a good disparity. If I can't blow your socks off I'll eat 'em! :)
Mike M.
@Mike - Yep sorry. I just tried with half a million rows and you were absolutely correct a couple of those additional operators start becoming important and CPU time goes through the roof.
Martin Smith
+3  A: 

Here's the pivot version:

http://odata.stackexchange.com/stackoverflow/q/7305/so3241450

-- SO3241450

CREATE TABLE #SO3241450 (
    Weekno int NOT NULL
    ,DayOfWeek int NOT NULL
    ,FromTime time NOT NULL
    ,ToTime time NOT NULL
)

INSERT INTO #SO3241450 VALUES
(1, 2, '10:00', '14:00')
,(1, 3, '10:00', '14:00')
,(2, 3, '08:00', '13:00')
,(2, 4, '09:00', '13:00')
,(2, 5, '14:00', '22:00')
,(3, 1, '06:00', '13:00')
,(3, 4, '06:00', '13:00')
,(3, 5, '14:00', '22:00')

;WITH Base AS (
    SELECT Weekno, DayOfWeek, FromTime AS [Start], ToTime AS [End]
    FROM #SO3241450
)
,norm AS (
SELECT Weekno, ColName + CONVERT(varchar, DayOfWeek) AS ColName, ColValue
FROM Base
UNPIVOT (ColValue FOR ColName IN ([Start], [End])) AS pvt
)
SELECT *
FROM norm
PIVOT (MIN(ColValue) FOR ColName IN ([Start1], [End1], [Start2], [End2], [Start3], [End3], [Start4], [End4], [Start5], [End5], [Start6], [End6], [Start7], [End7])) AS pvt​
Cade Roux
+1 for showing the pivot version
Mike M.