SubSelect or Joins? Is there a better way to write this mysql query?

Question

I know there is always a better way to do something, but I'm not sure how? What is the best way to optimize this query? Should I use Joins, a separate query, etc.. I know it's not a complex query.. just trying to expand my knowledge.

Any suggestions would be appreciated!

SELECT
  community_threads.id AS thread_id,
  community_threads.title AS thread_title,
  community_threads.date AS thread_date,
  community_threads.author_id AS author_id,
  `user`.display_name AS author_name,
  `user`.organization AS author_organization,
  (SELECT date FROM community_replies replies WHERE replies.thread_id = community_threads.id ORDER BY date DESC LIMIT 1) AS reply_date,
  (SELECT   count(id) FROM community_replies replies WHERE replies.thread_id = community_threads.id ORDER BY date DESC LIMIT 1) AS total_replies
FROM
  community_threads
INNER JOIN `user` ON community_threads.author_id = `user`.id
WHERE
  category_id = '1'
ORDER BY
  reply_date DESC
LIMIT 0, 5

Answer 1

This can be improved with a JOIN against a subselect which gets the aggregate COUNT() per thread_id and the aggregate MAX(date). Instead of evaluating the subselect for each row, the derived tables should be evaluated only once for the entire query and joined against the rest of the rows from community_threads.

SELECT
  community_threads.id AS thread_id,
  community_threads.title AS thread_title,
  community_threads.date AS thread_date,
  community_threads.author_id AS author_id,
  `user`.display_name AS author_name,
  `user`.organization AS author_organization,
  /* From the joined subqueries */
  maxdate.date AS reply_date,
  threadcount.num AS total_replies
FROM
  community_threads
  INNER JOIN `user` ON community_threads.author_id = `user`.id
  /* JOIN against subqueries to return MAX(date) (same as order by date DESC limit 1) and COUNT(*) from replies */
  /* number of replies per thread_id */
  INNER JOIN  (
    SELECT thread_id, COUNT(*) AS num FROM replies GROUP BY thread_id
  ) threadcount ON community_threads.id = threadcount.thread_id
  /* Most recent date per thread_id */
  INNER JOIN (
    SELECT thread_id, MAX(date) AS date FROM replies GROUP BY thread_id
  ) maxdate ON community_threads.id = maxdate.thread_id
WHERE
  category_id = '1'
ORDER BY
  reply_date DESC
LIMIT 0, 5

You may get even better performance if you put the LIMIT 0, 5 inside the reply_date subquery. That will only pull the most recent 5 in the subquery, and the INNER JOIN will discard all from community_threads not matching.

/* I *think* this will work...*/
SELECT
  community_threads.id AS thread_id,
  community_threads.title AS thread_title,
  community_threads.date AS thread_date,
  community_threads.author_id AS author_id,
  `user`.display_name AS author_name,
  `user`.organization AS author_organization,
  /* From the joined subqueries */
  maxdate.date AS reply_date,
  threadcount.num AS total_replies
FROM
  community_threads
  INNER JOIN `user` ON community_threads.author_id = `user`.id
  INNER JOIN  (
    SELECT thread_id, COUNT(*) AS num FROM replies GROUP BY thread_id
  ) threadcount ON community_threads.id = threadcount.thread_id
  /* LIMIT in this subquery */
  INNER JOIN (
    SELECT thread_id, MAX(date) AS date FROM replies GROUP BY thread_id ORDER BY date DESC LIMIT 0, 5
  ) maxdate ON community_threads.id = maxdate.thread_id
WHERE
  category_id = '1'
ORDER BY
  reply_date DESC

Answer 2

Far as I can tell, this looks like a good opportunity for a group by.

SELECT 
    community_threads.id AS thread_id,
    community_threads.title AS thread_title,
    community_threads.date AS thread_date,
    community_threads.author_id AS author_id,
    `user`.display_name AS author_name,
    `user`.organization AS author_organization,
    MAX(replies.date) AS reply_date,
    COUNT(replies.id) AS total_replies
FROM community_threads
INNER JOIN `user` ON community_threads.author_id = `user`.id
INNER JOIN community_replies AS replies ON replies.thread_id = community_threads.id
WHERE category_id = 1
GROUP BY thread_id, thread_title, thread_date, author_id, author_name, author_organization)
ORDER BY reply_date DESC
LIMIT 0, 5

Hope this helps.