Question

I have multiple applications each with unique application ID’s, Each application has logs which are stored in a table. I want to know how to calculate the date difference
eg:

    AppID Start           Loggedon   change 
    A1    08/07/2010      08/09/2010 Xchange
    A1    08/07/2010      08/20/2010 Ychange
    A1    08/07/2010      08/30/2010 Zchange
    A2    07/07/2010      07/13/2010 Ychange
    A3    09/07/2010      09/09/2010 Xchange

So I want the values

Difference
2   (Difference between the application start date and 1st loggedon date)
11  (difference between 08/20 and 08/09, the prior row because AppID stayed the same)
10  (difference between 08/30 and 08/20, the prior row because AppID stayed the same)
6   (Difference between the application start date and 1st loggedon date)
2   (Difference between the application start date and 1st loggedon date)

Hope I am clear. How can I achieve this, I tried Ranking and Row_Number. But I might be wrong somewhere. I am using SQL Server and cannot use LAG()

Answer 1

@Hogan has the correct approach, but for some reason, I can’t get it to work completely. Here is a slight variation that seems to produce the correct result — however, please accept @Hogan’s answer as he beat me to it 🙂

WITH cte AS (
  SELECT AppId, Start, LoggedOn, 
  ROW_NUMBER() OVER (ORDER BY AppId) rn
FROM Apps
  ) 
SELECT c.AppId,
  CASE 
    WHEN c2.RN IS NULL
    THEN DATEDIFF(day,c.start,c.Loggedon)
    ELSE
         DATEDIFF(day,c2.Loggedon,c.Loggedon)
    END as TimeWanted
FROM cte c
   LEFT JOIN cte c2 on c.AppId = c2.AppId
      AND c.rn = c2.rn + 1

And here is the Fiddle.

Good luck.