Question

I’m developing a little rss feed emailer.

The four tables i have are Users, Feeds, Subscriptions, and Distribution. Feeds contain the actual feed details, Subscriptions connect the Users with the Feeds, and Distribution records the emails transactions of the rss feed posts.

I’m trying to query the Distribution table to get the latest record for each user/feed subscription. I have the following query but it obviously only returns one row. I need to use a subquery but I can’t wrap my head around how.

SELECT d.id, d.user_id, d.feed_id, d.created
FROM Distribution AS d
INNER JOIN Feeds AS f ON f.id = d.feed_id
INNER JOIN Subscriptions AS s ON s.feed_id = d.feed_id
GROUP BY d.id, d.user_id, d.feed_id
ORDER BY d.created DESC 
LIMIT 1

Distribution table data
id, created, feed_id, post_id, user_id, success

(0, '2012-08-31 09:37:49', 20, 3, 2, 1)
(1, '2012-08-25 09:36:21', 20, 1, 1, 1)
(2, '2012-08-25 09:37:49', 21, 1, 2, 1)
(4, '2012-08-25 09:39:06', 21, 4, 1, 1)
(5, '2012-08-25 10:12:29', 20, 7, 2, 0)
(6, '2011-05-24 10:34:30', 20, 112, 1, 0)

the following query produces the following results

SELECT Distribution.*
FROM   Distribution NATURAL JOIN (
  SELECT   user_id, feed_id, MAX(created) AS created
  FROM     Distribution
  GROUP BY user_id, feed_id
) t

(0, '2012-08-31 09:37:49', 20, 3, 2, 1)
(1, '2012-08-25 09:36:21', 20, 1, 1, 1)
(2, '2012-08-25 09:37:49', 21, 1, 2, 1)
(4, '2012-08-25 09:39:06', 21, 4, 1, 1)

Answer 1

As explained in the MySQL manual:

The server is free to choose any value from each group, so unless they are the same, the values chosen are indeterminate. Furthermore, the selection of values from each group cannot be influenced by adding an ORDER BY clause. Sorting of the result set occurs after values have been chosen, and ORDER BY does not affect which values the server chooses.

You are trying to find the groupwise maximum, which requires use of a subquery to identify the latest record.

The subquery itself should use MySQL’s MAX() function on the created column to identify the corresponding value of most recent record within each group, then use that information to join against the table in the outer/parent query. I think this is what you’re after, but it’s hard to be certain without your table schema/sample data (at very least it should set you on the right path):

SELECT Distribution.*
FROM   Distribution NATURAL JOIN (
  SELECT   user_id, feed_id, MAX(created) AS created
  FROM     Distribution
  GROUP BY user_id, feed_id
) t