Mysql SELECT distinct Group by and Order by - mysql

Question

I have problem with my SQL SELECT statement. I get in right order, right drivers, but my other columns are incorrect! And I can't get it right way.
I have data like this:
id, races_id, drivers_id, drive_nr, lap_nr, time, dnf
"231", "9", "41", "1", "1", "00:00:04.750", "0"
"232", "9", "41", "1", "2", "00:00:06.030", "0"
"233", "9", "41", "1", "3", "00:00:01.740", "0"
"234", "9", "42", "1", "1", "00:00:05.440", "0"
"235", "9", "42", "1", "2", "00:00:05.400", "0"
"236", "9", "42", "1", "3", "00:00:02.300", "0"
"237", "9", "43", "1", "1", "00:00:00.620", "0"
"238", "9", "43", "1", "2", "00:00:00.290", "0"
"239", "9", "43", "1", "3", "00:00:00.280", "0"
"240", "9", "44", "1", "1", "00:00:00.600", "0"
"241", "9", "44", "1", "2", "00:00:00.190", "0"
"242", "9", "44", "1", "3", "00:00:00.220", "0"
"243", "9", "45", "1", "1", "00:00:02.830", "0"
"244", "9", "45", "1", "2", "00:00:01.890", "0"
"245", "9", "45", "1", "3", "00:00:03.200", "0"
"246", "9", "46", "1", "1", "00:00:03.580", "0"
"247", "9", "46", "1", "2", "00:00:04.550", "0"
"248", "9", "46", "1", "3", "00:00:01.060", "0"
"249", "9", "47", "1", "1", "00:00:02.920", "0"
"250", "9", "47", "1", "2", "00:00:03.950", "0"
"251", "9", "47", "1", "3", "00:00:00.320", "0"
"252", "9", "48", "1", "1", "00:00:02.150", "0"
"253", "9", "48", "1", "2", "00:00:05.720", "0"
"254", "9", "48", "1", "3", "00:00:04.530", "0"
"255", "9", "49", "1", "1", "00:00:01.530", "0"
"256", "9", "49", "1", "2", "00:00:04.360", "0"
"257", "9", "49", "1", "3", "00:00:07.110", "0"
"258", "9", "50", "1", "1", "00:00:00.450", "0"
"259", "9", "50", "1", "2", "00:00:03.550", "0"
"260", "9", "50", "1", "3", "00:00:07.900", "0"
with query this:
SELECT `id` ,
`races_id` ,
`drivers_id` ,
`drive_nr` ,
`lap_nr` ,
MIN( `time` ) AS TIME,
`dnf`
FROM `laps`
WHERE `races_id` =9
GROUP BY drivers_id`
ORDER BY MIN( `time` ) ASC
I get:
id, races_id, drivers_id, drive_nr, lap_nr, time, dnf
240, 9, 44, 1, 1, 00:00:00.190, 0
237, 9, 43, 1, 1, 00:00:00.280, 0
249, 9, 47, 1, 1, 00:00:00.320, 0
258, 9, 50, 1, 1, 00:00:00.450, 0
246, 9, 46, 1, 1, 00:00:01.060, 0
255, 9, 49, 1, 1, 00:00:01.530, 0
231, 9, 41, 1, 1, 00:00:01.740, 0
243, 9, 45, 1, 1, 00:00:01.890, 0
252, 9, 48, 1, 1, 00:00:02.150, 0
234, 9, 42, 1, 1, 00:00:02.300, 0
so I get correct time column in correct order, but not others columns like ID, drive_nr, lap_nr, dnf
how to fix my query to get distinct drivers_id with min time with correct other data?

Answer 1

And if you remove "GROUP BY"
SELECT `id` , `races_id` , `drivers_id` , `drive_nr` , `lap_nr` , MIN( `time` ) AS TIME, `dnf` FROM `laps` WHERE `races_id` = 9 ORDER BY MIN( `time` ) ASC
Take a look on this link, use of group by and min

Answer 2

The problem is that GROUP BY groups rows together for an aggregate function. In standard SQL every column returned must be a mentioned in the group by clause or an aggregate field, but MySQL extends this.
However although MySQL does allow extra columns to be returned, it does not specify which row the value of those columns comes from. While there is a pattern (seems to be the last row inserted I think), this is not defined and could change.
To get the other fields you have a couple of options.
Simplest is to have a sub query that gets the driver id and the min time for a lap for that driver id, then join that back against the laps table (joining on the driver id and the time) top get the values of the others fields for the matching row. There are a couple of minor downs sides to this. Firstly MySQL will not use an index on the fields on the sub query to join to the main table, but with limited data probably not an issue (beyond the annoyance of the queries popping up in the slow query log). The 2nd issue is if someone has a best lap time shared between 2 laps.
Simple example of the sql:-
SELECT a.id,
a.races_id,
a.drivers_id,
a.drive_nr,
a.lap_nr,
mt.min_time,
a.dnf
FROM laps a
INNER JOIN
(
SELECT drivers_id ,
MIN( `time` ) AS min_time
FROM laps
WHERE races_id = 9
GROUP BY drivers_id
) mt
ON a.drivers_id = mt.drivers_id
AND a.`time` = mt.min_time
WHERE a.races_id = 9
ORDER BY min_time ASC
If you do have 2 laps with the same min lap time then you need to specify which ones details to return (or you might not care, and could just misuse GROUP BY on the outer query as well).
A second solution is that you generate a sequence number for each row on the results ordered by driver id and lap time, resetting the sequence number on change of driver id. Then discard any lap which does not have a sequence of 1. However harder to read, and likely to be very slow when you have lots of data.
Example as follows (not tested):-
SELECT id,
races_id,
drivers_id,
drive_nr,
lap_nr,
`time`,
dnf
FROM
(
SELECT id,
races_id,
drivers_id,
drive_nr,
lap_nr,
`time`,
dnf,
#ctr := IF(drivers_id = #did, #ctr + 1, 1) AS ctr,
#did := drivers_id
FROM
(
SELECT id,
races_id,
drivers_id,
drive_nr,
lap_nr,
`time`,
dnf
FROM laps
WHERE races_id = 9
ORDER BY drivers_id, `time`
)
CROSS JOIN
(
SELECT #ctr := 0, #did := 0
) sub1
) sub2
WHERE ctr = 1
ORDER BY `time`