Question

Assuming I have two tables, “users” and “audit”, and the audit table is populated by a trigger each time the user’s status changes. I’m trying to do a select from both tables, and get the user entry (unique) with the associated latest audit entry. Say the tables have the following fields:

users:

• id
• username
• email
• status

audit:

• id (just an arbitrary autoincrement)
• user_id (fk)
• new_status
• created_at (date/time when this happened)

Thus, for each user entry, there’s multiple audit entries.

So, eventually I need the result to be say (where the italic part is the results from audit, and the non-italic is the user part of the row in the mysql result-set):
1, john, [e-mail], 10, 5, 1, 10, 2012-08-10-10:15:59
etc

Now, the only way I can see is to do compounded selects, where I do something like:
SELECT U.*, (SELECT * FROM audit A2 where A2.id = max(A.id)) FROM users U left join audit A on (U.id = A.user_id);

Problem is, if you’re having thousands or perhaps hundreds of thousands of results in user, you’ll be doing n+1 queries internally in the DB – which is really inefficient.

Any ideas on how to write this as one SQL statement without using compounded SELECTS?

Answer 1

I am assuming there is always a corresponding audit entry, and have thus used INNER JOIN. Also assuming there are not two audit entries with the same created_at date for the same user.

select u.id, u.username, u.email, u.status, 
    a.id, a.new_status, a.created_at
from users u
inner join (
    select user_id, max(created_at) as MaxCreated
    from audit
    group by user_id
) am on u.id = am.user_id
inner join audit a on am.user_id = a.user_id and am.MaxCreated = a.created_at