From: | Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: planner doesn't use bitmap index |
Date: | 2015-10-29 18:32:44 |
Message-ID: | CAFj8pRAcZxK2FOmpxGvyrkFMBa81nMRRBXEuahrFVXA0uDgJBQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
2015-10-29 19:20 GMT+01:00 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> writes:
> > -- I was surprised, so following query can use index
> > postgres=# explain select a from test2 where a at time zone
> > 'America/Santiago' >= now() at time zone 'America/Santiago' ;
> > QUERY
> > PLAN
> >
> ══════════════════════════════════════════════════════════════════════════════════════════════
> > Index Only Scan using test2_a_idx on test2 (cost=0.13..12.18 rows=1
> > width=8)
> > Filter: (timezone('America/Santiago'::text, a) >=
> > timezone('America/Santiago'::text, now()))
> > (2 rows)
>
> This plan isn't actually "using" the index in any meaningful way; it's
> applying the where condition as a filter. It happens to be sane to use
> the index as a dumb data source, because it can be an index-only scan, and
> that might (if you're lucky and don't hit too many recheckable rows) be
> cheaper than a seqscan. But we don't consider plain indexscans as worth
> the trouble to consider in such cases, because a full-table plain
> indexscan can never beat a seqscan, either in the planner's cost model or
> in reality.
>
> > why, the index isn't used in this case?
> > postgres=# explain select a,b from test2 where a at time zone
> > 'America/Santiago' >= now() at time zone 'America/Santiago' ;
>
> Can't be an index-only scan because of the use of b, so there's no
> possible way that this can be better than a seqscan.
>
I understand.
Than you for explanation.
Regards
Pavel
>
> regards, tom lane
>
From | Date | Subject | |
---|---|---|---|
Next Message | Pavel Stehule | 2015-10-29 19:24:52 | Re: proposal: PL/Pythonu - function ereport |
Previous Message | Tom Lane | 2015-10-29 18:25:14 | Re: ALTER ... OWNER TO ... vs. ALTER DEFAULT PRIVILEGES |