Question

SELECT User.username,
       IFNULL(SUM(credit) - SUM(debit), 0) AS total,
       ( SELECT SUM(Bid.credit) - SUM(Bid.debit) AS freebids
           FROM users AS User
           LEFT
           JOIN bids AS Bid
             ON Bid.user_id = User.id
          WHERE Bid.type = 2
       ) AS FreeBids,
       ( SELECT SUM(Bid.credit) - SUM(Bid.debit) AS Normalbids
           FROM users AS User
           LEFT
           JOIN bids AS Bid
             ON Bid.user_id = User.id
          WHERE Bid.type = 0
             OR Bid.type = 1
       ) AS NormalBids
  FROM users AS User
  LEFT
  JOIN bids AS Bid
    ON Bid.user_id = User.id
 GROUP
    BY User.id

here is my sample table:
for users

id username 
1  user1
2  user2
3  user3
4  user4

For bids

id user_id debit credit type
1  1       0     10     0
2  1       1     5      2 
3  1       1     0      2
4  3       0     10     0
6  2       0     10     0
7  4       1     10     0
8  4       1     0      1

But I am having problem in displaying the subquery (Freebids and Normalbids) are all has the same value here are the sample display:

username    total   FreeBids    NormalBids
user1       10       12809          965
user2       20       12809          965
user3       9        12809          965
user4       0        12809          965

I can’t figure out where did I go worng in my query.
Any suggestion to solve my problem?
Thank you for your help.

Answer 1

The reason why all the same values showed up was because there was no distinction between the users table in the outer query and the users table in the inner query. They both have the same alias. You must distinguish either the outer users table or the inner users table by assigning it with a different alias (such as u) so that the subquery knows which table to reference for the outer value:

... FROM users AS u ...

Having said that, it is highly inefficient to use subqueries like that as they would need to be executed for each row in your users table (we’re talking about entire tables needing to be joined and filtered as many times as there are users x2).

Your query could be rewritten MUCH more efficiently using conditional aggregation:

SELECT
    a.username,
    IFNULL(SUM(b.credit) - SUM(b.debit), 0) AS total,
    SUM(IF(b.type = 2, b.credit, 0)) - SUM(IF(b.type = 2, b.debit, 0)) AS freebids,
    SUM(IF(b.type IN (0,1), b.credit, 0)) - SUM(IF(b.type IN (0,1), b.debit, 0)) AS normalbids
FROM users a
LEFT JOIN bids b ON a.id = b.user_id
GROUP BY a.id, a.username

Which basically says: SUM the credit/debit ONLY if the type is x value, and you can SUM/COUNT/AVG/etc. on any condition you want just working with one table join.

---

SQL-Fiddle Demo