From: Rob Sargent <robjsargent@gmail.com>
To: matthias schoeneich <matthias.schoeneich@gmx.net>
Cc: pgsql-sql@postgresql.org
Subject: Re: Efficiency in Multiple Queries of Same Table in a PL/PgSQL Function
Date: Sat, 17 Oct 2009 23:26:56 -0600
Message-ID: <4ADAA720.3050706@gmail.com> (raw)
In-Reply-To: <hbc9qk$vc4$1@news.hub.org>
References: <302670f20910161927h688e13e7qd096e63e68b7ae6c@mail.gmail.com>
<4AD93837.7030500@gmail.com>
<hbc9qk$vc4$1@news.hub.org>
But if I read the OP correctly the sigma are in fact used additively in
each row in blah. "sigma_* = sigma_* +"
matthias schoeneich wrote:
> Hi,
>
> as you don't seem to need the sigma_*'s, you could calc the whole
> result with one query using:
>
> CREATE OR REPLACE FUNCTION poly_example2() RETURNS SETOF FLOAT8 AS
> $poly_example$
> DECLARE
> f_result FLOAT8 := 0.0;
> i_rowcount INT := 0 ;
>
> BEGIN
> SELECT sum((RANDOM() * 100 ) * (term.i * term.i) + RANDOM() *
> (term.j * term.j) + term.k) +
> sum((RANDOM() * 53 ) * (term.i * term.i) +(RANDOM()* 5) *
> (term.j * term.j) + term.k) +
> sum( 96.232234 * (term.i * term.i) + 0.32322325 *
> (term.j * term.j) + term.k) ,
> count(*)
> INTO f_result ,
> i_rowcount
> FROM blah AS term;
>
> IF i_rowcount > 0 THEN
> RETURN NEXT f_result;
> ELSE
> RETURN NEXT 0;
> END IF;
> END;
> $poly_example$ LANGUAGE plpgsql STRICT IMMUTABLE;
>
> I've just put it in your plpgsql body to handle the case where table
> blah contains no rows.
>
> Matthias
>
> Rob Sargent schrieb:
>> I don't see anything in the assignment statements (sigma_* :=) which
>> would prevent one from doing all three of them within a single for
>> loop. In fact, written as is there's some chance the values of the
>> sigma_*s might change between repeated calls to the function since
>> there is no explicit ordering of the rows returned from table blah.
>> Putting all the assignments into a single select from blah would at
>> least say that the sigma values are from the same dataset per run.
>>
>>
>> As to efficiency in general, I would expect the entire table (~50
>> rows) would be entirely in memory after the first select, but you
>> plan triples the time in the loop. This expense would likely only be
>> noticeable if the function itself is called /lots/.
>>
>> Gary Chambers wrote:
>>> All...
>>>
>>> In the poly_example function below, I am emulating an actual
>>> requirement by querying the same table three (3) times in order to
>>> derive a solution to a problem. Is this the best or most efficient
>>> and effective way to implement this? The table (which consists of
>>> only five (5) FLOAT8 columns) I'm querying contains less than 50 rows.
>>> Thanks in advance for any insight or criticisms you offer.
>>>
>>> CREATE OR REPLACE FUNCTION poly_example() RETURNS SETOF FLOAT8 AS
>>> $poly_example$
>>> DECLARE
>>> term blah%ROWTYPE;
>>> sigma_l FLOAT8 := 0.0;
>>> sigma_b FLOAT8 := 0.0;
>>> sigma_r FLOAT8 := 0.0;
>>>
>>> BEGIN
>>> FOR term in SELECT * FROM blah LOOP
>>> sigma_l := sigma_l + (RANDOM() * 100) * (term.i * term.i) +
>>> RANDOM() * (term.j * term.j) + term.k;
>>> END LOOP;
>>>
>>> FOR term in SELECT * FROM blah LOOP
>>> sigma_b := sigma_b + (RANDOM() * 53) * (term.i * term.i) +
>>> (RANDOM() * 5) * (term.j * term.j) + term.k;
>>> END LOOP;
>>>
>>> FOR term in SELECT * FROM blah LOOP
>>> sigma_r := sigma_r + 96.232234 * (term.i * term.i) +
>>> 0.32322325 * (term.j * term.j) + term.k;
>>> END LOOP;
>>>
>>> RETURN NEXT sigma_l + sigma_b + sigma_r;
>>> END;
>>> $poly_example$ LANGUAGE plpgsql STRICT IMMUTABLE;
>>>
>>> -- Gary Chambers
>>>
>>> /* Nothing fancy and nothing Microsoft! */
>>>
>>>
>>
>>
>
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: robjsargent@gmail.com, matthias.schoeneich@gmx.net
Subject: Re: Efficiency in Multiple Queries of Same Table in a PL/PgSQL Function
In-Reply-To: <4ADAA720.3050706@gmail.com>
* 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