MySQL returning other columns with DISTINCT - mysql

Question

Need a little bit of help with a query.
SELECT id,email FROM user_info WHERE username!=''
AND email='example#gmail.com'
GROUP BY email ORDER BY id DESC LIMIT 1
What this does currently is fetch a distinct email but not the newest ID for an account under this email.

Answer 1

It sounds like you can apply an aggregate to the id field and it will return the most recent id for each email:
SELECT Max(id), email
FROM user_info
WHERE username!=''
AND email='example#gmail.com'
GROUP BY email
Unfortunately, when you apply a GROUP BY without an aggregate there is no guarantee what id will be returned unless you specify to select the max.
If you want to return the username that is associated with the max(id), then you can use a subquery:
SELECT i.MaxId,
i.email,
u.username
FROM user_info u
inner join
(
select max(id) MaxId, email
from user_info
WHERE username!=''
AND email='example#gmail.com'
group by email
) i
on u.id = i.maxid
and u.email = i.email
WHERE username!=''
AND email='example#gmail.com';

Answer 2

If by "newest id" you mean the largest value, then here's one approach that makes use of MySQL user variables to retain the value from previous rows, so a comparison can be made:
SELECT IF(u.email=#prev_email,1,0) AS dup_email_ind
, u.id
, u.username
, #prev_email := u.email AS email
FROM user_info u
CROSS
JOIN ( SELECT #prev_email := NULL) i
WHERE u.username != ''
AND u.email = 'example#gmail.com'
GROUP
BY u.email DESC
, u.id DESC
, u.username DESC
This returns all of the rows, with an indicator of whether the row is considered an older "duplicate" email or not. Rows that have dup_email_ind = 1 are identified as older duplicates, dup_email_ind = 0 indicates that this row is the latest row (the row with the largest id value) for a given email value.
(Usually, when I'm looking for duplicates like this, it's helpful for me to return both, or all, of the rows that are "duplicates".)
To return only the rows with the "newest id", wrap the query above (as an inline view) in another query: the output from the query is used a row source for the outer query.)
SELECT d.*
FROM (
-- the query above gets put here
) d
WHERE d.dup_mail_ind = 0
Another approach is to use a correlated subqueries in the SELECT list, although this is really only suitable for returning small sets. (This approach can have serious performance issues with large sets.)
SELECT ( SELECT u1.id
FROM user_info u1
WHERE u1.email = e.email
AND u1.username != ''
ORDER BY u1.id DESC
LIMIT 1
) AS id
, ( SELECT u2.username
FROM user_info u2
WHERE u2.email = e.email
AND u2.username != ''
ORDER BY u2.id DESC
LIMIT 1
) AS username
, e.email
FROM ( SELECT u.email
FROM user_info u
WHERE u.email = 'example#gmail.com'
AND u.username != ''
GROUP BY u.email
) e