The Table:
(`post_id`, `forum_id`, `topic_id`, `post_time`)
(79, 8, 4, '2012-11-19 06:58:08');
(80, 3, 3, '2012-11-19 06:58:42'),
(81, 9, 9, '2012-11-19 06:59:04'),
(82, 11, 6, '2012-11-19 16:05:39'),
(83, 9, 9, '2012-11-19 16:07:46'),
(84, 9, 11, '2012-11-19 16:09:33'),
The Query:
SELECT post_id, forum_id, topic_id FROM posts
GROUP BY topic_id
ORDER BY post_time DESC
LIMIT 5
The Results:
[0] => [post_id] => 84 [forum_id] => 9 [topic_id] => 11
[1] => [post_id] => 82 [forum_id] => 11 [topic_id] => 6
[2] => [post_id] => 81 [forum_id] => 9 [topic_id] => 9
[3] => [post_id] => 80 [forum_id] => 3 [topic_id] => 3
[4] => [post_id] => 79 [forum_id] => 8 [topic_id] => 4
The Problem:
How to rewrite the query so that it returns post_id -> 83 instead of post_id -> 81 ?
They both have the same forum and topic ids, but post_id -> 81 has an older date than post_id -> 83.
But it seems that Group By gets the 'first' record and not the 'newest' one.
I tried changing the query to
SELECT post_id, forum_id, topic_id, MAX(post_time)
but that returns both post_id 81 and 83
If you select attributes that are not used in the group clause, and are not aggregates, the result is unspecified. I.e you don't know which rows the other attributes are selected from. (The sql standard does not allow such queries, but MySQL is more relaxed).
The query should then be written e.g. as
SELECT post_id, forum_id, topic_id
FROM posts p
WHERE post_time =
(SELECT max(post_time) FROM posts p2
WHERE p2.topic_id = p.topic_id
AND p2.forum_id = p.forum_id)
GROUP BY forum_id, topic_id, post_id
ORDER BY post_time DESC
LIMIT 5;
or
SELECT post_id, forum_id, topic_id FROM posts
NATURAL JOIN
(SELECT forum_id, topic_id, max(post_time) AS post_time
FROM posts
GROUP BY forum_id, topic_id) p
ORDER BY post_time
LIMIT 5;
It's not very beautiful , but it works:
SELECT * FROM (SELECT post_id, forum_id, topic_id FROM posts
ORDER BY post_time DESC) as temp
GROUP BY topic_id
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With