Combine Multiple Rows into One Row with data separated into two columns

Question

I have a sql Table of the following manner:

Table_Name  Field_Name       New_Value  Create_Date     
TableTest   Loading Hours       25      12/12/2015  2:56:26 p.m.    
TableTest   Loading Hours       15      12/12/2015  2:30:23 p.m.    
TableTest   Loading Hours       11      12/12/2015  2:09:42 p.m.    
TableTest   Loading_Percentage  35.8    12/12/2015  2:56:26 p.m.    
TableTest   Loading_Percentage  15.5    12/12/2015  2:30:23 p.m.    

What i want to do is to be able to transform the above, How best can I do this efficiently? I could potentially have more values in the loading hours in the table above, and this would need to appear logically as presented in the transformed data below.

TableTest   Field_Name         OldValue NewValue   DateTime 
TableTest   Loading Hours       15       25        12/12/2015   2:56:26 p.m.
TableTest   Loading Hours       11       15        12/12/2015   2:30:23 p.m.
TableTest   Loading Hours       NULL     11        12/12/2015   2:09:42 p.m.
TableTest   Loading_Percentage  15.5     35.8      12/12/2015   2:56:26 p.m.
TableTest   Loading_Percentage  NULL     15.5      12/12/2015   2:30:23 p.m.

Thanks for the help in advance.

Cheers

Answer 1

Your scenario is exactly what the lag window function is designed for. Unfortunately, lag is not available in SQL Server 2008. Until you upgrade, you can do the following:

with cte as (
  select *,
         row_number() over (partition by table_name, field_name order by create_date) as rn
    from tbl
)
select curr.table_name,
       curr.field_name,
       prev.new_value as old_value,
       curr.new_value,
       curr.create_date
  from cte curr
  left join cte prev
    on prev.table_name = curr.table_name
   and prev.field_name = curr.field_name
    on prev.rn = curr.rn + 1
 order by curr.table_name, curr.field_name, curr.rn desc

If you ever upgrade to SQL Server 2012+, the query can be written more simply, and get a nice boost in performance, by using lag :

select table_name,
       field_name,
       lag(new_value) over (partition by table_name, field_name order by create_date) as old_value,
       new_value,
       create_date
  from tbl
 order by table_name, field_name, create_date desc

Answer 2

Handle this scenario in Trigger. Add Insert and Update trigger for the table and use the trigger to achieve the functionality which you required. When ever insert/update the data into the table, the trigger should get the last inserted/updated value based on the Field_Name and insert a new record with New and old value.