Received: from localhost (postgresql.org [64.49.215.8])
	by postgresql.org (Postfix) with ESMTP id 3DCEC475D0F
	for <pgsql-sql@postgresql.org>; Mon, 13 Jan 2003 05:39:34 -0500 (EST)
Received: from serwer.skawsoft.com.pl (serwer.skawsoft.com.pl [213.25.37.66])
	by postgresql.org (Postfix) with ESMTP id 65DC2475C9E
	for <pgsql-sql@postgresql.org>; Mon, 13 Jan 2003 05:39:33 -0500 (EST)
Received: from klaster.net (pe210.krakow.cvx.ppp.tpnet.pl [213.76.40.210])
	by serwer.skawsoft.com.pl (Postfix) with ESMTP
	id 4DA262B21E; Mon, 13 Jan 2003 11:36:51 +0100 (CET)
Message-ID: <3E229858.4070509@klaster.net>
Date: Mon, 13 Jan 2003 11:43:36 +0100
From: Tomasz Myrta <jasiek@klaster.net>
User-Agent: Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US;
	rv:1.2b) Gecko/20021016
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: =?ISO-8859-1?Q?Jimmy_M=E4kel=E4?= <jimmy.makela@agent25.se>
Cc: "'pgsql-sql@postgresql.org'" <pgsql-sql@postgresql.org>
Subject: Re: Unique indexes not unique?
References: <D1045567F50DD311AB1B00508B3188E9026546D8@RINGHALS>
In-Reply-To: <D1045567F50DD311AB1B00508B3188E9026546D8@RINGHALS>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
X-Virus-Scanned: by AMaViS new-20020517
X-Archive-Number: 200301/164
X-Sequence-Number: 11029

Jimmy Mäkelä wrote:

> I found that Postgres isn't behaving like I thought when using a 
> unique index in
> combination with NULL-values...
> Is this a bug or specified in the SQL-standard? If its a bug, is it 
> fixed in a
> recent version? We are using 7.2.3
>
> This is the results I got:
>
> intranet=# create table foo (a varchar(10), b varchar(10));
> CREATE
> intranet=# create unique index foo_idx on foo using btree(a, b);
> CREATE
> intranet=# insert into "foo" (a, b) values ('apa', 'banan');
> INSERT 26229704 1
> intranet=# insert into "foo" (a, b) values ('apa', 'banan');
> ERROR:  Cannot insert a duplicate key into unique index foo_idx
> intranet=# insert into "foo" (a, b) values ('apa', null);
> INSERT 26229706 1
> intranet=# insert into "foo" (a, b) values ('apa', null);
> INSERT 26229707 1

I'm not sure unique index works properly for null values. I can't 
explain, why. Maybe it comes from SQL standard - null i a special value 
and can't be compared using default operators to other non null values:
1>null =null
1<null =null
1=null =null

>
>
> And another completely unrelated question... I have got a table with a 
> composite
> index on A andBb and an index on A
> which I query with something like this:
>
> SELECT * FROM "table"
> WHERE (a = 1 OR a = 2 OR a = 3) AND b > 1232132 AND b < 123123123213123
>
> Postgres then chooses to use the index for A three times, which is 
> really slow
> on my table...
> Then I rewrote the query like:
>
> SELECT * FROM "table"
> WHERE a = 1 AND b > 1232132 AND b < 123123123213123
> UNION SELECT * FROM "table"
> WHERE a = 2 AND b > 1232132 AND b < 123123123213123
> UNION SELECT * FROM "table"
> WHERE a = 3 AND b > 1232132 AND b < 123123123213123


Try to rewrite your query to show postgres how to use index on AB:
SELECT * FROM "table"
WHERE
(a = 1 AND b > 1232132 AND b < 123123123213123) or
(a = 2 AND b > 1232132 AND b < 123123123213123) or
(a = 3 AND b > 1232132 AND b < 123123123213123);

Regards,
Tomasz Myrta