Getting wrong values for other columns when I select MAX(updated_date)

问题

I am using select field1, max(updated_date) from mytable.
I get the correct value for max(updated_date), i.e. the largest date.
However for field1 I just get the the value for the first record, i.e. "ta1" when I really want the "ta3" value from the third record (the one with the max date value).

e.g.

+------------+---------------------+
| field1     | update_date         |
+------------+---------------------+
| ta1        | 2012-03-11 11:05:15 |
| ta2        | 2012-03-11 11:05:32 |
| ta3        | 2012-03-11 11:05:56 |
+------------+---------------------+
3 rows in set (0.00 sec)

+------------+---------------------+
| field1     | max(update_date)    |
+------------+---------------------+
| ta1        | 2012-03-11 11:05:56 |
+------------+---------------------+
1 row in set (0.00 sec)

回答1:

You either need a GROUP BY clause or a more complex query.

SELECT field1, MAX(updated_date)
  FROM mytable
 GROUP BY field1

For the sample data, this will return 3 rows.

More likely, you want:

SELECT t1.field1, t3.max_date
  FROM mytable AS t1
  JOIN (SELECT MAX(t2.updated_date) AS max_date
          FROM mytable AS t2
       ) AS t3
    ON t1.updated_date = t3.max_date;

For the sample data, this will return 1 row:

ta3   2012-03-11 11:05:56

Of the major DBMS, only MySQL allows you to omit the GROUP BY clause when you have a mixture of aggregates and non-aggregate columns in the select-list. The SQL standard requires the GROUP BY clause and you must list all non-aggregate columns in it. Sometimes, in MySQL, omitting the GROUP BY clause produces the answer you want; as often as not, though, it manages to give an unexpected answer.

回答2:

Use ORDER BY and LIMIT.
Then it's as simple as:

  SELECT field1, updated_date 
    FROM mytable
ORDER BY updated_date DESC 
   LIMIT 1;

If the query is needed a lot you can try this alternative:

  SELECT t1.field1, t1.updated_date 
    FROM mytable t1
         LEFT JOIN mytable t2 
     AND t2.updated_date > t1.updated_date
   WHERE t2.field1 IS NULL;

Short explanation:
For each row, give me any rows with a more-recent updated_date.
But (WHERE clause) take only the row with no more-recent updated_date.
The technique is sometimes called a self-exclusion join.
This is the intermediate result (without WHERE clause, and adding t2.* to SELECT list):

ta1    2012-03-11 11:05:15    ta2     2012-03-11 11:05:32
ta1    2012-03-11 11:05:15    ta3     2012-03-11 11:05:56
ta2    2012-03-11 11:05:32    ta3     2012-03-11 11:05:56
ta3    2012-03-11 11:05:56    null    null

来源：https://stackoverflow.com/questions/9657695/getting-wrong-values-for-other-columns-when-i-select-maxupdated-date

标签

mysql

sql

postgresql

rdbms

max