Attempt at work around of int4 query won't touch int8 index ... - Mailing list pgsql-performance

From James Robinson
Subject Attempt at work around of int4 query won't touch int8 index ...
Date
Msg-id 2756862E-E3F5-11D7-8513-000A9566A412@socialserve.com
Whole thread Raw
Responses Re: Attempt at work around of int4 query won't touch int8 index ...
List pgsql-performance
Hello,

    I'm trying a work-around on the "index on int8 column gets ignored by
planner when queried by literal numbers lacking the explicit '::int8'"
issue, and had hoped that perhaps I could create a functional index on
the result of casting the pk field to int4, and mabye with a little
luck the planner would consider the functional index instead. Here's
what I'm playing with on 7.3.4:

social=# create table foo (id int8 primary key, stuff text);
NOTICE:  CREATE TABLE / PRIMARY KEY will create implicit index
'foo_pkey' for table 'foo'
CREATE TABLE
social=# create index foo_pkey_int4 on foo(int4(id));
CREATE INDEX

social=# explain analyze select id from foo where id = 42;
                                          QUERY PLAN
------------------------------------------------------------------------
--------------------
  Seq Scan on foo  (cost=0.00..22.50 rows=1 width=8) (actual
time=0.01..0.01 rows=0 loops=1)
    Filter: (id = 42)
  Total runtime: 0.15 msec
(3 rows)

social=# explain analyze select id from foo where id = 42::int8;
                                                  QUERY PLAN
------------------------------------------------------------------------
------------------------------------
  Index Scan using foo_pkey on foo  (cost=0.00..4.82 rows=1 width=8)
(actual time=0.02..0.02 rows=0 loops=1)
    Index Cond: (id = 42::bigint)
  Total runtime: 0.09 msec
(3 rows)

social=# explain analyze select id from foo where id = int4(33);
                                          QUERY PLAN
------------------------------------------------------------------------
--------------------
  Seq Scan on foo  (cost=0.00..22.50 rows=1 width=8) (actual
time=0.01..0.01 rows=0 loops=1)
    Filter: (id = 33)
  Total runtime: 0.07 msec
(3 rows)

Is this just a dead end, or is there some variation of this that might
possibly work, so that ultimately an undoctored literal number, when
applied to an int8 column, could find an index?

Thanks,
James


pgsql-performance by date:

Previous
From: Mary Edie Meredith
Date:
Subject: Re: [osdldbt-general] Re: [GENERAL] how to get accurate
Next
From: Tom Lane
Date:
Subject: Re: Attempt at work around of int4 query won't touch int8 index ...