Tom,
I've tried like you said (::integer) but doesn't work. I tried another
way, using GROUP BY at the end, but no luck again. What now?
Table "prog_teste"
Attribute | Type | Modifier
-----------------+--------------+----------
cod_teste | integer | not null
...
Index: xpkprog_teste
Index "xpkprog_teste"
Attribute | Type
-----------+---------
cod_teste | integer
unique btree (primary key)
sitest=# explain select * from prog_teste where cod_teste=90::integer;
NOTICE: QUERY PLAN:
Seq Scan on prog_teste (cost=0.00..1.21 rows=1 width=138)
EXPLAIN
sitest=# explain select * from prog_teste where cod_teste=90::integer order
by cod_teste;
NOTICE: QUERY PLAN:
Sort (cost=1.22..1.22 rows=1 width=138)
-> Seq Scan on prog_teste (cost=0.00..1.21 rows=1 width=138)
Best regards,
José Vilson de Mello de Farias
Dígitro Tecnologia - Brasil
----- Original Message -----
From: Tom Lane <[EMAIL PROTECTED]>
To: Vilson farias <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>; SIMONE Carla MOSENA
<[EMAIL PROTECTED]>
Sent: Sexta-feira, 10 de Novembro de 2000 13:11
Subject: Re: [GENERAL] Sequencial scan over primary keys
: "Vilson farias" <[EMAIL PROTECTED]> writes:
: > sitest=# CREATE TABLE tipo_categoria (
: > sitest(# cod_categoria smallint NOT NULL,
: > sitest(# descricao varchar(40),
: > sitest(# CONSTRAINT XPKtipo_categoria PRIMARY KEY (cod_categoria)
: > sitest(#
: > sitest(# );
:
: > sitest=# explain select * from tipo_categoria where cod_categoria = 1;
: > NOTICE: QUERY PLAN:
:
: > Seq Scan on tipo_categoria (cost=0.00..22.50 rows=10 width=14)
:
: Try it with
: select * from tipo_categoria where cod_categoria = 1::smallint;
:
: An unadorned literal "1" is an int, not a smallint, and the planner is
: not currently very smart about indexing cross-datatype comparisons.
:
: Alternatively, just declare your table with column type int. Because of
: alignment requirements for the varchar column, you're not actually
: saving any space by using the smallint declaration anyway.
:
: regards, tom lane
:
