Get sum of a column in a join: derived table

In the example below I wish to return a row for each row in b and return the
sum of a.val where b.id = a.id. At the bottom is a working select but I don't
know how it works!  I learnt the technique from a friend but it is a
mystery to me and I don't know if there are any better techniques for
achieving my goal.

I have read in the Manual (section 4.2.1.x) references to "derived
tables". Clearly a2 in my example is derived. What are the rules for
derivation; can they come only after a FROM? The section in the manual
simply states this as and "interesting case!".

How do you go about optimising this sort of query?

----------------------------------------------------------------------------
4.2.1.3. Subqueries

    Subqueries specifying a derived table must be enclosed in
    parentheses and must be assigned a table alias name. (See Section
    4.2.1.2.) For example:

    FROM (SELECT * FROM table1) AS alias_name

    This example is equivalent to FROM table1 AS alias_name. More
    interesting cases, which can't be reduced to a plain join, arise
    when the subquery involves grouping or aggregation.

----------------------------------------------------------------------------

dd=# select * from b;         dd=# select * from a;
 id | val                      id | val
----+-----                    ----+-----
  2 |   1                       1 |   2
  4 |   1                       2 |   4
(2 rows)                        3 |   3
                                2 |   5
                              (4 rows)

----------------------------------------------------------------------------

dd=#
    SELECT
        b.id as bid, b.val as bval, a2.id, sum as aval
    FROM b
        LEFT OUTER JOIN
            (select id, sum(val) from a group by id) as a2
            ON b.id = a2.id;

 bid | bval | id | aval
-----+------+----+------
   2 |    1 |  2 |    9
   4 |    1 |    |
(2 rows)

--
Rory Campbell-Lange
<rory@campbell-lange.net>
<www.campbell-lange.net>