Using distinct keyword with join

Question

My professor has given me an assignment.

Write a query that will produce the snum values of all salespeople (suppress the duplicates) with orders in the Orders table.

SalesPeople

Snum number(4)
Sname varchar2(10)
City varchar2(10)
Comm number(3,2)

Customer

Cnum number(4)
Cname varchar2(10)
City varchar2(10)
Rating number(4)
Snum number(4)

Orders

Onum number(4)
Amt number(7,2)
Odate date
Cnum number(4)
Snum number(4)

I am not sure if I have understood the question completely.

I have written the query using join.

select distinct s.snum,onum
from salespeople s, ordrs o
where s.snum = o.snum
order by snum;

and the output is

      SNUM       ONUM
---------- ----------
      1001       3003
      1001       3008
      1001       3011
      1002       3005
      1002       3007
      1002       3010
      1004       3002
      1007       3001
      1007       3006

But I don't want SNUM to be repeated. Can someone point me in a right direction ?

Thanks.

Answer 1

The question asks you to "Write a query that will produce the snum values of all salespeople (suppress the duplicates) with orders in the Orders table" - it doesn't specify that you need to include the order numbers. (It implies that you don't need the order numbers, since by including them you will duplicate the snum values.)

Given that snum appears on the Orders table, it should be relatively simple to SELECT a DISTINCT list of SNUM values FROM the ORDERS table.

Answer 2

In your query remove onum from the SELECT -

select distinct s.snum
from salespeople s, ordrs o
where s.snum = o.snum
order by snum;

This should display only the distinct snum.

Answer 3

You only need to select from 1 table:

select distinct o.snum
from ordrs o
order by o.snum;

NB I have kept the order by clause, but it isn't strictly needed to answer the question as it was given.

Answer 4

Both answer above seems correct but it seems that

select distinct snum
from ordrs
order by snum

or

select snum
from ordrs
group by snum
order by snum

would be more efficient because we dont need to read the salespeople. This is acceptable if there is a foreign key constraint on snum in the table ordrs.

NB: Tony just answered when i was writing this. I would 100% agree with him :)

Answer 5

And it can also be easily done with a subquery instead of a join:

select s.snum
from salespeople s
where exists (
  select 1
  from ordrs
  where snum = s.snum
);

which basically says: give me all snum's that have at least 1 matching record in the ordrs table.