Basic usage of SQL group

Question

A very basic question if this can be done using Group by in SQL Server.

I have a table like this:

ID   Version
-------------
123  1
123  3
158  1
158  2
158  4

using Group By ID, max(Version) I get

ID    Version
--------------
123   3
158   4

I have an extended table with another column value , which is the interesting data:

ID   Version   Value
----------------------
123  1         abc
123  3         xyz
158  1         pq
158  2         je
158  4         kju

I want to retrieve

ID   Version   Value
----------------------
123  3         xyz
158  4         kju

I fail to get the value as shown above. Is this possible by using Group By ?

Answer 1

You can solve that with an INNER JOIN with a SUBQUERY. I don't know your table names but lets call them VersionTable and VersionValueTable. So I would do this:

SELECT vt.ID, vt.Version, vvt.Value
FROM VersionValueTable vvt
   INNER JOIN (SELECT ID, MAX(Version)
               FROM VERSION
               GROUP BY ID) vt ON vt.ID = vvt.ID AND vt.Version = vvt.Version

Answer 2

You don't use group by for this. The most common approach is row_number() :

select t.*
from (select t.*,
             row_number() over (partition by id order by version desc) as seqnum
      from t
     ) t
where seqnum = 1;

There are a zillion other ways to do this. If you are learning about window functions, the closest to the group by is:

select t.*
from (select t.*,
             max(verson) over (partition by id) as max_version
      from t
     ) t
where version = max_version;

Answer 3

You can compare version of rows with Max-version that belongs to each row ID . With a subquery you can find the subset of rows that have the same ID and get the maximum value of Version field. As you see in the following code :

SELECT * FROM MyTable t1
    WHERE t1.version = 
      (SELECT max(version) FROM MyTable t2 WHERE t2.id = t1.id)

To find a subset of records, that have the same id as the current row in main Select query , you can use a whereclause ( WHERE t2.id = t1.id ) where t1 is alias for our table in main query and t2 is the alias for the table in subquery.