Received: from astercity.net (root@astercity.net [195.116.8.78])
	by hub.org (8.9.1/8.9.1) with ESMTP id RAA09941
	for <pgsql-sql@postgreSQL.org>; Mon, 11 Jan 1999 17:48:47 -0500 (EST)
	(envelope-from pawel@astercity.net)
Received: from mail.astercity.net (verona58.waw.pdi.net [195.116.8.58])
	by astercity.net (8.9.2/8.9.2) with SMTP id XAA00213
	for <pgsql-sql@postgreSQL.org>; Mon, 11 Jan 1999 23:48:09 +0100 (EET)
Date: Mon, 11 Jan 1999 23:39:11 +0100
From: Pawel Pierscionek <pawel@astercity.net>
X-Mailer: The Bat! (v1.19) S/N 359CE73E
Reply-To: Pawel Pierscionek <pawel@astercity.net>
Organization: PDi- Publiczny Dostep do Internetu
X-Priority: 3 (Normal)
Message-ID: <19985.990111@astercity.net>
To: pgsql-sql <pgsql-sql@postgreSQL.org>
Subject: Temporary tables
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Hi,

I have to play a lot with temp tables.
Could you pls, give me a hint on emulating them under Postgres.

I have to use them quite often and with 10000 rows each.
I  don't use indices cause I only cache a search result and don't need
any sorting, searching on temptables.

What's faster:

1)
once:
 create tmptable;

many times:
 delete from tmptable;
 insert into tmptable select ... ;

or

2)
many times:
drop tmptable;
select ... into tmptable;

I  prefer  the  second way since I have a lot of concurrent access and
create unique named tables under each transaction.
Unfortunately the DROP thing dumps a core under PLPGSQL function. :(

TIA,
Pawel