HOW TO: SQL Server select distinct field based on max value in other field

问题

id   tmpname      date_used        tkt_nr
---|---------|------------------|--------|
1  | template| 04/03/2009 16:10 | 00011  |
2  | templat1| 04/03/2009 16:11 | 00011  |
5  | templat2| 04/03/2009 16:12 | 00011  |
3  | diffname| 03/03/2009 15:11 | 00022  |
4  | diffname| 03/03/2009 16:12 | 00022  |
6  | another | 03/03/2009 16:13 | NULL   |
7  | somethin| 24/12/2008 11:12 | 00023  |
8  | name    | 01/01/2009 12:12 | 00026  |

I would like to have the result:

id   tmpname      date_used        tkt_nr
---|---------|------------------|--------|
5  | templat2| 04/03/2009 16:12 | 00011  |
4  | diffname| 03/03/2009 16:12 | 00022  |
7  | somethin| 24/12/2008 11:12 | 00023  |
8  | name    | 01/01/2009 12:12 | 00026  |

So what I'm looking for is to have distinct tkt_nr values excluding NULL, based on the max value of datetime.

I have tried several options but always failed

SELECT *
FROM  templateFeedback a 
JOIN (
      SELECT ticket_number, MAX(date_used) date_used
      FROM   templateFeedback
      GROUP BY ticket_number
     ) b 
ON a.ticket_number = b.ticket_number AND a.date_used = b.date_used

I would appreciate any help. Unfortunately I need the code to be compatible with SQL Server.

回答1:

I've stopped doing things this way since I discovered windowing functions. Too often, there are two records with the same timestamp and I get two records in the resultset. Here's the code for tSQL. Similar for Oracle. I don't think mySQL supports this yet.

Select id, tmpname, date_used, tkt_nbr
From 
(
    Select id, tmpname, date_used, tkt_nbr,
        rownum = Row_Number() Over (Partition by tkt_nbr Order by date_used desc)
) x
Where row_num=1

来源：https://stackoverflow.com/questions/12570747/how-to-sql-server-select-distinct-field-based-on-max-value-in-other-field

标签

sql

sql-server

distinct

max