Question

Input

ID  RowID   Data
1   1   S
1   1   Q
1   1   L
1   1   Null
1   1   Null
1   1   S
1   1   E
1   1   R
1   1   V
1   1   E
1   1   R
1   1   Null
1   1   DB
1   2   S
1   2   T
1   2   A
1   2   C
1   2   K
2   1   O
2   1   V
2   1   E
2   1   R
2   1   Null
2   1   Null
2   1   Null
2   1   F
2   1   L
2   1   O
2   1   W

Expected Output

ID  NewData
1   SQL2SERVER1DB,STACK
2   OVER3FLOW

Answer 1

The only way to guarantee the order of the output is to add an column which defines the correct order to the table – I’ve used an IDENTITYcolumn.

I’ve also assumed that there is a mistake in your sample data creation script (see my comment on the question) – the revised sample data creation script is as follows:

DECLARE @t TABLE(ID INT IDENTITY(1,1) , groupId  INT, RowID INT, Data VARCHAR(10))
INSERT INTO @t (groupId, RowID,DATA)
SELECT 1,1,'S' UNION ALL SELECT 1,1,'Q' UNION ALL SELECT 1,1,'L' UNION ALL
SELECT 1,1,NULL UNION ALL SELECT 1,1,NULL UNION ALL SELECT 1,1,'S' UNION ALL
SELECT 1,1,'E' UNION ALL SELECT 1,1,'R' UNION ALL SELECT 1,1,'V' UNION ALL SELECT 1,1,'E' UNION ALL
SELECT 1,1,'R' UNION ALL SELECT 1,1,NULL UNION ALL SELECT 1,1,'DB' UNION ALL
SELECT 1,2,'S' UNION ALL SELECT 1,2,'T' UNION ALL SELECT 1,2,'A' UNION ALL SELECT 1,2,'C' UNION ALL
SELECT 1,2,'K' UNION ALL SELECT 2,1,'O' UNION ALL SELECT 2,1,'V' UNION ALL SELECT 2,1,'E' UNION ALL 
SELECT 2,1,'R'  UNION ALL SELECT 2,1,NULL UNION ALL SELECT 2,1,NULL UNION ALL SELECT 2,1,NULL
UNION ALL SELECT 2,1,'F' UNION ALL SELECT 2,1,'L' UNION ALL SELECT 2,1,'O' UNION ALL SELECT 2,1,'W'

It’s not possible, as far as I can see, to correctly aggregate the NULL rows and concatenate the strings in a single step, so this solution uses chained CTEs to carry out the steps one at a time. I’ve also used a CTE to deduplicate the output (rather than grouping).

--convert consecutive nulls to a count
;WITH nullCTE
AS
(
    SELECT ID, ID AS e
    FROM @t AS t
    WHERE t.Data IS NULL
    AND EXISTS (SELECT 1 FROM @t AS q WHERE q.Data IS NOT NULL AND q.Id = t.ID + 1)

    UNION ALL

    SELECT  c.ID, t.Id
    FROM    @t AS t
    JOIN    nullCTE AS c
    ON      t.ID = c.e - 1
    WHERE t.Data IS NULL
)
-- simplify the null table
,grpCTE
AS
(SELECT ID, ID - e + 1 SIZE,
    ROW_NUMBER() OVER (PARTITION BY ID
                       ORDER BY e
                      ) AS rn
 FROM nullCTE
)
--build the output
,outputCTE
AS
(
    SELECT  t1.groupId,
            (
                SELECT COALESCE(t.Data,CAST(g.size AS VARCHAR(11)),'') + CASE WHEN u.RowID > t.RowID AND u.groupId = t.groupId THEN ',' ELSE '' END
                FROM        @t AS t
                LEFT JOIN   grpCTE  AS g
                ON          g.ID = t.ID
                AND         rn = 1
                LEFT JOIN   @t AS u
                ON          u.ID = t.Id + 1
                WHERE t.groupID = t1.groupId
                ORDER BY t.Id
                FOR XML PATH('') 
            ) AS NewData,
            ROW_NUMBER() OVER (PARTITION BY groupId
                               ORDER BY Id
                              ) AS rn
    FROM @t AS t1
)
SELECT groupId, NewData
FROM outputCTE
WHERE rn = 1