The following bug has been logged on the website:
Bug reference: 12871
Logged by: Maksym Boguk
Email address: maxim.boguk@gmail.com
PostgreSQL version: 9.4.1
Operating system: Linux
Description:
As my bugreport BUG #12862 stalled in queue for unknown reason I'll try
post it again.
First test case showing issue:
postgres=# create database test;
CREATE DATABASE
postgres=# \c test
You are now connected to database "test" as user "postgres".
test=# create table test as select array[100]::integer[] as f1 from
generate_series(1,10000);
SELECT 10000
test=# analyze test;
ANALYZE
test=# explain analyze select * from test where f1 && array[100];
QUERY PLAN
---------------------------------------------------------------------------------------------------------
Seq Scan on test (cost=0.00..532.40 rows=10000 width=25) (actual
time=0.048..6.207 rows=10000 loops=1)
Filter: (f1 && '{100}'::integer[])
Total runtime: 7.154 ms
(3 rows)
test=# create extension intarray;
CREATE EXTENSION
test=# explain analyze select * from test where f1 && array[100];
QUERY PLAN
------------------------------------------------------------------------------------------------------
Seq Scan on test (cost=0.00..532.40 rows=10 width=25) (actual
time=0.062..5.598 rows=10000 loops=1)
Filter: (f1 && '{100}'::integer[])
Total runtime: 6.548 ms
(3 rows)
As you could see after installing the intarray extension PostgreSQL started
use a general selectivity estimator. Estimated rows=10000 before create
extension and estimated rows=10 after, 10000 - correct one.
Problem that custom (redefined) intarray && @> <@ operators are defined
with
contsel/contjoinsel instead of arraycontsel/arraycontjoinsel.
However, simple change extension/intarray--1.0.sql with correct CREATE
OPERATOR doesn't fix issue:
test=# explain analyze select * from test where f1 && array[100];
ERROR: arraycontsel called for unrecognized operator 814221170
I cannot find any quick fix for this issue because OID of custom &&
operator
isn't fixed, so no simple change of arraycontsel/arraycontjoinsel possible.
An issue exists in any version since introducing
arraycontsel/arraycontjoinsel in 9.2 up to HEAD.
