Pair Row data into column data - sql-server

Question

I have a query that returns data in the form
[SetID],[COLID],[DataValue],[Order].
1,1,'Value1',1
1,1,'Value2',2
1,1,'Value3',3
1,2,'value4',1
1,2,'Value5',2
1,2,'Value6',3
How to make it return the data like
[SetID],'Value1','Value4',1
[SetID],'Value2','Value5',2
[SetID],'Value3','Value6',3
So that the data tagged with each columnid is back into a column.
The answer appears to be some sort of pivot, but I'm not looking for a straight conversion, or a one-row answer.

Answer 1

To be able to put each data value into a separate column, you need to know the number of DataValues and the ColId of each. You can't dynamically return variable number of columns (unless you use dynamic SQL to construct the query on the fly), moreover each row needs to have the same number of columns, otherwise the table structure doesn't make sense.
Here's an example that you can quickly test, it uses an in-memory table.
Given the table MyTable:
declare #MyTable table (SetId int, ColId int, DataValue varchar(20), [Order] int);
insert into #MyTable values (1,1,'Value1',1)
insert into #MyTable values (1,1,'Value2',2)
insert into #MyTable values (1,1,'Value3',3)
insert into #MyTable values (1,2,'Value4',1)
insert into #MyTable values (1,2,'Value5',2)
insert into #MyTable values (1,2,'Value6',3)
This query will return the values in separate columns:
select mt.SetId,
( select top 1 m1.DataValue from #MyTable m1
where m1.SetId = mt.SetId and
m1.[Order] = mt.[Order] and
m1.ColId = 1 -- hard-coded ColId, you need to know its value
) 'val1',
( select top 1 m1.DataValue from #MyTable m1
where m1.SetId = mt.SetId and
m1.[Order] = mt.[Order] and
m1.ColId = 2 -- same as above
) 'val2', mt.[Order]
from #MyTable mt
group by mt.SetId, mt.[Order]
Result:
+-----------------------------------+
| SetId | Val1 | Val2 | Order |
|-------+---------+--------+--------|
| 1 | Value1 | Value4 | 1 |
|-------+---------+--------+--------|
| 1 | Value2 | Value5 | 2 |
|-------+---------+--------+--------|
| 1 | Value3 | Value6 | 3 |
+-----------------------------------+
However, I would recommend stuffing all the data values into one column (separated by a delimiter), because this will work with any number of data values. Here's the query:
select mt.SetId,
stuff(( select ', '+ m2.DataValue
from #MyTable m2
where mt.SetId = m2.SetId and
mt.[Order] = m2.[Order]
FOR XML PATH('')) , 1, 2, '') 'DataValues',
mt.[Order]
from #MyTable mt
group by mt.SetId, mt.[Order]
Result
+----------------------------------+
| SetId | DataValues | Order |
|-------+------------------+-------|
| 1 | Value1, Value4 | 1 |
|-------+------------------+-------|
| 1 | Value2, Value5 | 2 |
|-------+------------------+-------|
| 1 | Value3, Value6 | 3 |
+-------+------------------+-------+