Question

I have a MySQL (5.1.49) table with three columns.

mysql> create table testme(id int auto_increment, id2 int not null, somedate datetime, primary key(id));

In my case id2 is not unique, but I want to return rows with distinct id2 values with the max somedate.

Here’s some sample data.

mysql> insert into testme values (1, 5, '2012-01-02 01:01:01'),(2, 5, '2012-02-02 02:02:02'),(3, 7, '2010-01-10 11:01:33');

This question almost answers mine, but with the extra id field, the returned id and id2 don’t match. For id2=5, it’s returning id=1 instead of id=2.

mysql> select id, id2, max(somedate) from testme group by id2;
+----+-----+---------------------+
| id | id2 | max(somedate)       |
+----+-----+---------------------+
|  1 |   5 | 2012-02-02 02:02:02 |
|  3 |   7 | 2010-01-10 11:01:33 |
+----+-----+---------------------+

I’m expecting,

+----+-----+---------------------+
| id | id2 | max(somedate)       |
+----+-----+---------------------+
|  2 |   5 | 2012-02-02 02:02:02 |
|  3 |   7 | 2010-01-10 11:01:33 |
+----+-----+---------------------+

Want the ID that matches maximum date for each ID2

Does anyone have any ideas please? Thanks

Answer 1

This query would definitely work, although it may be not the optimal one:

select t.id, s.id2, s.somedate 
from testme t 
  join 
    ( select id2, max(somedate) as somedate 
      from testme 
      group by id2
    ) s 
    on  s.id2 = t.id2 
    and s.somedate = t.somedate;