-- SQL -- GROUP BY clause: non-aggregate fields mandate

Question

Hello,
I was studying Databases, (particularly the retrieval of the data), and found something interesting.
While using an Aggregate Function in the SELECT clause, it is mandatory to have all the non-aggregate fields in the SELECT clause to be there in the GROUP BY clause.
For example,
SELECT dept_no, SUM(salary)
FROM employee
GROUP BY dept_no;
The above SQL works fine.
But, what if the user misses the dept_no in the GROUP BY clause or he/she misses the GROUP BY clause itself?
Certainly, it is an error.
Why is this error not handled by the database. I mean, the database should be smart/intelligent enough to add the GROUP BY clause by itself. So suppose, if I miss out the GROUP BY clause or miss a non-aggregate field from the SELECT clause when I am having at least one aggregate function on a field with at least one non-aggregated field in the SELECT clause, the database should check the GROUP BY clause at time of compilation and add the mandate missed out fields in the GROUP BY clause.
Example,
SQL1:_
SELECT dept_no, SUM(salary)
FROM employee
GROUP BY dept_no;
SQL2:_
SELECT dept_no, SUM(salary)
FROM employee;
Here, the SQL1 and SQL2, both should give me same outputs without an error.
I am unable to understand why is this not handled?

Answer 1

Hi,
998478 wrote:
... If we mix aggregate and non-aggregate values then there must be a GROUP BY clause containing all the non-aggregate values. Why is this not handled by the database/compiler itself? It IS handled by the compiler itself. The compiler handles it by raising an error. The compiler has no way of knowing whether you want to remove something from the SELECT clause, or to add something to the GROUP BY clause, or not to use aggregate functions, or to use more aggregate functions, or some combination of the above. If the compiler re-wrote your code, and did any of these things automatically, it would be wrong more often than it was right, and you would (rightly) be complaining about its behavior.
For example, this is clearly wrong:
SELECT    deptno
,       job
,       SUM (sal)
FROM       scott.emp
GROUP BY  deptno
;What is the right way to fix it?
<h3>1. Remove something from the SELECT clause</h3>
SELECT    deptno
,       SUM (sal)
FROM       scott.emp
GROUP BY  deptno
;<h3>2. Add something to the GROUP BY clause</h3>
SELECT    deptno
,       job
,       SUM (sal)
FROM       scott.emp
GROUP BY  deptno
,         job
;<h3>3. Not use aggregate functions</h3>
SELECT    deptno
,       job
,       sal
FROM       scott.emp
;<h3>4. Use more aggregate functions</h3>
SELECT    deptno
,       MIN (job)
,       SUM (sal)
FROM       scott.emp
GROUP BY  deptno
;These aren't all the options, either. For example, the correct fix might be to use analytic functions instead of aggregate functions.
How can anybody say which of these is right? All of them are the right answer for some problem.
By the way, saying that everying in the SELECT clause must be an aggregate or in the GROUP BY clause is a bit over-simplified.
More completely, here are the ABC's of GROUP BY:
When you use a GROUP BY clause and/or an aggregate function, then everything in the SELECT clause must be:
(A) an <b>A</b>ggregate function,
(B) one of the "group <b>B</b>y" expressions,
(C) a <b>C</b>onstant, or
(D) something that <b>D</b>epends entirely on the above. (For example, if you "GROUP BY TRUNC(dt)", you can SELECT "TO_CHAR (TRUNC(dt), 'Mon-DD')").
Edited by: Frank Kulash on Apr 13, 2013 1:44 PM
Added code examples.