[jablan]

Evo skripta za kreiranje tabele:

CREATE TABLE items (
    a character varying(10),
    b character varying(10),
    total integer,
    id integer NOT NULL
);

INSERT INTO items (a, b, total, id) VALUES ('a', 'b', 10, 1);
INSERT INTO items (a, b, total, id) VALUES ('a', 'c', 10, 2);
INSERT INTO items (a, b, total, id) VALUES ('a', 'd', 10, 3);
INSERT INTO items (a, b, total, id) VALUES ('b', 'd', 10, 4);
INSERT INTO items (a, b, total, id) VALUES ('b', 'd', 10, 5);
INSERT INTO items (a, b, total, id) VALUES ('c', 'd', 10, 6);
INSERT INTO items (a, b, total, id) VALUES ('c', 'c', 10, 7);
INSERT INTO items (a, b, total, id) VALUES ('c', 'b', 10, 8);
INSERT INTO items (a, b, total, id) VALUES ('c', 'd', 10, 9);
INSERT INTO items (a, b, total, id) VALUES ('c', 'b', 10, 10);
INSERT INTO items (a, b, total, id) VALUES ('a', 'b', 10, 11);

I upit:

test=# select a, b, sum(total) from items group by a, b;
 a | b | sum 
---+---+-----
 b | d |  20
 a | c |  10
 c | b |  20
 a | d |  10
 c | d |  20
 a | b |  20
 c | c |  10
(7 rows)

Elem, meni treba upit koji vraća COUNT(*) za prethodni upit (dakle 7), ali bez korišćenja podupita, tj. bez da se ceo gornji upit wrapuje u SELECT COUNT(*) FROM ...