Question

I am using a group by clause in my query. I want to get other columns not specified in the group by parameters

SELECT un.user, un.role 
FROM [Unique] un 
group by user, role

In the query about [Unique] has 7 columns altogether. How do I get the other columns?

Answer 1

In most databases (MySQL and SQLite are the exceptions I know of), you cannot include a column in a GROUP BY SELECT unless:

1. The column is included in the GROUP BY clause.

2. The column is aggregated in one of the supported aggregate functions.

In MySQL and SQLite, the rows inside the aggregate groups from which the extra values get taken are undefined.

If you want extra columns in any other engine, you can wrap the column names in MAX():

 SELECT un.user, un.role, MAX(un.city), MAX(un.bday)
 FROM [Unique] un 
 GROUP BY user, role

In this case, the values for the extra columns are likely to come from different rows in the input record set. If this is important (sometimes it isn’t since the extra columns come from the one side of a one-to-many JOIN), you can’t use this technique.

Just to be clear: If you use GROUP BY in a SELECT, then each row you get back is constructed out of groups of multiple rows in the table you’re SELECTing against. If you include columns that are not part of the GROUP BY clause, you’re not giving the engine any instructions on which row from the table you want that value read from. Most engines, therefore, do not allow you to run this kind of SQL. MySQL does, with undefined results but I personally consider it bad practice to do this.