DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=mime-version:in-reply-to:references:date:message-id:subject:from:to
	:cc:content-type:content-transfer-encoding;
	b=GfZmmnsBxRkb7A8Lk4/S3O3GyK37HdootLR5NFGzUUGqTa4c5y8zgi0QC3G0QSQBeJ
	8ZVVs/RpGs8esrMoEouL/dB8vyaMB/4esfW1rg67wHl9EWByoxlgdDaSLGCY3I5rX8yw
	YoYBwjQi8b38DR/SPz10nOr0/cte4BKxOYkkc=
MIME-Version: 1.0
In-Reply-To: <71F491F5DA99604A80DE49424BF3D02B0C7292EF@exchange8.actuate.com>
References: <71F491F5DA99604A80DE49424BF3D02B0C7290CA@exchange8.actuate.com>
	<20090811062503.86993bf6.darcy@druid.net>
	<162867790908110355g3b62cc56u1c74092dd01bbd8e@mail.gmail.com>
	<71F491F5DA99604A80DE49424BF3D02B0C7292EF@exchange8.actuate.com>
Date: Tue, 11 Aug 2009 19:22:55 +0200
Message-ID: <162867790908111022y48e65d31kcf5a037c1b41857f@mail.gmail.com>
Subject: Re: Updating a specific number of rows in pl/pgsql
From: Pavel Stehule <pavel.stehule@gmail.com>
To: Peter Headland <pheadland@actuate.com>
Cc: "D'Arcy J.M. Cain" <darcy@druid.net>, pgsql-sql@postgresql.org
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

2009/8/11 Peter Headland <pheadland@actuate.com>:
>> there are one fast trick
>> http://www.postgres.cz/index.php/PostgreSQL_SQL_Tricks#Fast_first_n_rows=
_removing
>
> Thanks - that's a very useful page!
>
> Unfortunately, there is no single column that provides a unique id, and I=
 am reluctant to add one (for example, using a sequence and a new index) fo=
r performance reasons.

ctid is unique system column in every table.

postgres=3D# create table x(a int);
CREATE TABLE
Time: 655,062 ms
postgres=3D# insert into x values(10);
INSERT 0 1
Time: 49,237 ms
postgres=3D# insert into x values(10);
INSERT 0 1
Time: 1,740 ms
postgres=3D# select ctid, a from x;
 ctid  | a
-------+----
 (0,1) | 10
 (0,2) | 10
(2 rows)


>
> Given that additional constraint, is my original plan using a loop to ite=
rate over a cursor reasonable? I don't anticipate p_number_of_items being m=
ore than 20.

why not? for small number of iteration is loop over cursor good solution.

Pavel Stehule

>
> --
> Peter Headland
> Architect
> Actuate Corporation
>
>
> -----Original Message-----
> From: Pavel Stehule [mailto:pavel.stehule@gmail.com]
> Sent: Tuesday, August 11, 2009 03:55
> To: D'Arcy J.M. Cain
> Cc: Peter Headland; pgsql-sql@postgresql.org
> Subject: Re: [SQL] Updating a specific number of rows in pl/pgsql
>
> 2009/8/11 D'Arcy J.M. Cain <darcy@druid.net>:
>> On Mon, 10 Aug 2009 17:52:36 -0700
>> "Peter Headland" <pheadland@actuate.com> wrote:
>>> I can get the rows I want to update like this:
>>>
>>> =C2=A0 SELECT *
>>> =C2=A0 =C2=A0FROM queue
>>> =C2=A0 =C2=A0WHERE id =3D p_queue_id
>>> =C2=A0 =C2=A0ORDER BY rank
>>> =C2=A0 =C2=A0LIMIT p_number_of_items;
>>>
>>> Of course, there may not be p_number_of_items available in the queue.
>>>
>>> I want to update all the rows in the cursor in the same way:
>>>
>>> =C2=A0 UPDATE queue SET assigned =3D TRUE;
>>
>> Assuming that there is a unique identifier on queue, let's call it
>> queue_id, you should be able to do something like this:
>>
>> =C2=A0UPDATE queue SET assigned =3D TRUE
>> =C2=A0WHERE queue_id IN (SELECT queue_id
>> =C2=A0 =C2=A0FROM queue
>> =C2=A0 =C2=A0WHERE id =3D p_queue_id
>> =C2=A0 =C2=A0ORDER BY rank
>> =C2=A0 =C2=A0LIMIT p_number_of_items);
>>
>
> there are one fast trick
> http://www.postgres.cz/index.php/PostgreSQL_SQL_Tricks#Fast_first_n_rows_=
removing
>
> p.s. replace DELETE by UPDATE
> regards
> Pavel Stehule
>
>> --
>> D'Arcy J.M. Cain <darcy@druid.net> =C2=A0 =C2=A0 =C2=A0 =C2=A0 | =C2=A0D=
emocracy is three wolves
>> http://www.druid.net/darcy/ =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =
=C2=A0 =C2=A0| =C2=A0and a sheep voting on
>> +1 416 425 1212 =C2=A0 =C2=A0 (DoD#0082) =C2=A0 =C2=A0(eNTP) =C2=A0 | =
=C2=A0what's for dinner.
>>
>> --
>> Sent via pgsql-sql mailing list (pgsql-sql@postgresql.org)
>> To make changes to your subscription:
>> http://www.postgresql.org/mailpref/pgsql-sql
>>
>