Question

Although I know this isn’t very efficient, I need to get this working fast so I try to follow the easiest path, or at least I thought so. I’m trying to get the max data out of an entity and take into consideration only the rows where USER is not null, and in case no rows have USER not set to null, then I will take the null ones into consideration.

The entity’s structure is as follows:

• Id int,
• ParentId int,
• Guid varchar,
• EventOn DateTime,
• User int (null)

So, the query is the following one:

select RS.[$Id] RangeSheet,
       GRST.EventOn,
       RSIS.Id [Status],
       RSIS.Name StatusRefex    
  from RangeSheet RS
left outer join (select RST.ParentId RangeSheet,
                        MAX(RST.EventOn) EventOn
                   from RangeSheetTime RST
                  where RST.[User] is (case RST.[User] when null then null else not null)
               group by RST.ParentId) GRST on RS.[$Id]=GRST.RangeSheet
left outer join RangeSheetTime RST on GRST.EventOn=RST.EventOn 
                                  and GRST.RangeSheet=RST.ParentId
left outer join RangeSheetItemState RSIS on (case when RST.StartOrEnd = 1 then 1 when RST.StartOrEnd = 0 then 4 else null end) = RSIS.Id
 where RS.[$IsDeleted]=0

The case I’m having trouble with is the one inside the view GRST.
What could I do to accomplish my requirements?

Answer 1

I’m a bit baffled about what you’re doing. Going off your text description though would something like the following help?

;WITH t AS
     ( SELECT  Id     ,
              ParentId,
              Guid    ,
              EventOn ,
              USER    ,
              RANK() OVER (PARTITION BY ParentId ORDER BY
              CASE
                       WHEN USER IS NULL
                       THEN 0
                       ELSE 1
              END DESC, EventOn DESC) Rnk
     )
SELECT *
FROM   t
WHERE  Rnk=1