From: Tom Lane <tgl@sss.pgh.pa.us>
To: Bryce Nesbitt <bryce2@obviously.com>
Cc: pgsql-sql@postgresql.org
Subject: Re: Sequential scan evaluating function for each row, seemingly needlessly
Date: Tue, 07 Sep 2010 16:31:39 -0400
Message-ID: <25196.1283891499@sss.pgh.pa.us> (raw)
In-Reply-To: <4C868F49.5050601@obviously.com>
References: <4C868F49.5050601@obviously.com>
Bryce Nesbitt <bryce2@obviously.com> writes:
> On psql 8.3.9, I ran a limited query limited to 5 results. There was a
> moderately expensive function call
> which I expected to be called 5 times, but was apparently called for
> each row of the sequential scan. Why?
Given the plan:
> Limit (cost=19654.53..19654.54 rows=5 width=12) (actual
> time=10001.976..10001.990 rows=5 loops=1)
> -> Sort (cost=19654.53..19826.16 rows=68651 width=12) (actual
> time=10001.972..10001.976 rows=5 loops=1)
> Sort Key: add_date
> Sort Method: top-N heapsort Memory: 25kB
> -> Seq Scan on extractq (cost=0.00..18514.26 rows=68651
> width=12) (actual time=19.145..9770.689 rows=73550 loops=1)
> Total runtime: 10002.150 ms
> (6 rows)
any interesting work is going to be done at the seqscan level. Sort
just sorts, and Limit just limits; neither do any user-defined
calculations. So yeah, your functions got run for every row of the
table. (This isn't totally a PG aberration, btw: if you read the SQL
spec closely you'll discover that ORDER BY is defined to happen after
any calculations specified in the SELECT list.)
You could try something like
select my_expensive_function(...), etc, etc from
(select * from some-tables order by foo limit n) ss;
where the inner select list just pulls the columns you'll need in
the outer calculations.
regards, tom lane
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-sql@postgresql.org
Cc: tgl@sss.pgh.pa.us, bryce2@obviously.com
Subject: Re: Sequential scan evaluating function for each row, seemingly needlessly
In-Reply-To: <25196.1283891499@sss.pgh.pa.us>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox