Question

I tried to run the following query:

select a,b,min(c) from tablename where e=1 and f=1 group by a,b order by a,b,c

But got the error:

Column name 'tablename.c' is invalid in the ORDER BY clause because it is not contained in either an aggregate function or the GROUP BY clause.

However c is contained in the aggregate function min(), so what’s the issue here? Is there a workaround?

This is on sqlserver 2000 (using the sqlserver 2008 management console).

Answer 1

As you are using the aggregation on that column, is no longer c but “aggregated c” that is why you can not use it in the order by clause.

You need to alias that column to use it in order by

select a,b,min(c) as min_c from tablename where e=1 and f=1 group by a,b order by a,b,min_c

EDIT:

Why the column c is not available ?

While in the statement we use clause group by, our original table i transformed into new temporary one.

In your case you used the column a and b to group, the the data in those column will not be changed only limited to unique.

On the column c you are using a function that for each unique group will retrieve the lowest value of c, for this purpose a new column has to be created that will store that results.

The order by clause is the last part of query that is executed over the select section. So over the result of the temporary table not the source one.

The simply order of statement execution goes like this:

FORM 
 JOIN
WHERE
GROUP BY 
HAVING 
SELECT 
ORDER BY