Learning the basics of EXISTS operator

Question

I am learning the basics of EXISTS operator.
create table loans
(acc_id number,
balance number(10,2));
insert into loans
  (acc_id, balance)
values
  (100, 20000);
insert into loans
  (acc_id, balance)
values
  (110, 22000);
insert into loans
  (acc_id, balance)
values
  (120, 7000);
insert into loans
  (acc_id, balance)
values
  (130, 172.99);
SQL> select * from loans;
    ACC_ID    BALANCE
       100      20000
       110      22000
       120       7000
       130     172.99
create table defaulters
  (cust_id number,
   name    varchar2(20),
   acc_id  number);
insert into defaulters
  (cust_id, name, acc_id)
values
  (1,'Vajaal',110);
insert into defaulters
  (cust_id, name, acc_id)
values
  (2,'Mostert',130);
SQL> select * from defaulters;
   CUST_ID NAME                     ACC_ID
         1 Vajaal                      110
         2 Mostert                     130
SQL> select acc_id from loans
  2  where exists(select 1 from defaulters
  3  where loans.acc_id=defaulters.acc_id);
    ACC_ID
       110
       130It just returns all acc_id rows in Loans table which has corresponding ACC_IDs present in defaulters.
On 10gR2 SQL Reference, for EXISTS operator, it says ' An EXISTS condition tests for existence of rows in a subquery' .
Would it be wrong if i say, EXISTS operator can be used when you want to return all rows in the Outerquery which has matching records in the Innerquery.

Answer 1

Would it be wrong if i say, EXISTS operator
can be used when you want to return all rows in the
Outerquery which has matching records in the
Innerquery.Depends on what you understand with "matching".
See following examples:
In this case matching means "<>" i.e. return all loans who have at least one row that differs in defaulters table (of course it returns all loans, because defaulters has 2 rows with different acc_id, BTW look also for NULLs and comparisons involving NULLs)
SQL> ed
Wrote file afiedt.buf
  1  select acc_id from loans
  2  where exists(
  3    select 1 from defaulters
  4*   where loans.acc_id<>defaulters.acc_id)
SQL> /
    ACC_ID
       100
       110
       120
       130See also following query:
SQL> ed
Wrote file afiedt.buf
  1  select acc_id from loans
  2  where exists(
  3    select 1 from defaulters
  4* )
SQL> /
    ACC_ID
       100
       110
       120
       130There isn't any condition at all, so any row in defaulters is "matching" row in this case.
So I'd stick with explanation in documentation this time ;)
Gints Plivna
http://www.gplivna.eu