path: root/doc/TODO.detail/rtree

From pgsql-bugs-owner+M10740=pgman=candle.pha.pa.us@postgresql.org Mon Jan 24 18:00:00 2005
Return-path: <pgsql-bugs-owner+M10740=pgman=candle.pha.pa.us@postgresql.org>
Received: from svr1.postgresql.org (svr1.postgresql.org [200.46.204.71])
	by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id j0ONxxw11019
	for <pgman@candle.pha.pa.us>; Mon, 24 Jan 2005 18:59:59 -0500 (EST)
Received: from localhost (unknown [200.46.204.144])
	by svr1.postgresql.org (Postfix) with ESMTP id 182703A4CB1
	for <pgman@candle.pha.pa.us>; Mon, 24 Jan 2005 23:59:55 +0000 (GMT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024)
	with ESMTP id 38013-02 for <pgman@candle.pha.pa.us>;
	Mon, 24 Jan 2005 23:59:53 +0000 (GMT)
Received: from postgresql.org (svr1.postgresql.org [200.46.204.71])
	by svr1.postgresql.org (Postfix) with ESMTP id 4B17E3A4B12
	for <pgman@candle.pha.pa.us>; Mon, 24 Jan 2005 23:59:54 +0000 (GMT)
X-Original-To: pgsql-bugs-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.144])
	by svr1.postgresql.org (Postfix) with ESMTP id E7E323A53D0
	for <pgsql-bugs-postgresql.org@localhost.postgresql.org>; Mon, 24 Jan 2005 23:44:58 +0000 (GMT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024)
	with ESMTP id 14037-08
	for <pgsql-bugs-postgresql.org@localhost.postgresql.org>;
	Mon, 24 Jan 2005 23:44:45 +0000 (GMT)
Received: by svr1.postgresql.org (Postfix, from userid 1001)
	id 1C2D53A585A; Tue, 25 Jan 2005 00:59:20 +0000 (GMT)
Received: from floppy.pyrenet.fr (floppy.pyrenet.fr [194.250.190.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C0CB23A573E
	for <pgsql-bugs@postgresql.org>; Mon, 24 Jan 2005 23:29:17 +0000 (GMT)
Received: by floppy.pyrenet.fr (Postfix, from userid 106)
	id EF14E31D91; Tue, 25 Jan 2005 00:29:16 +0100 (MET)
From: Andrew - Supernews <andrew+nonews@supernews.com>
X-Newsgroups: pgsql.bugs
Subject: [BUGS] incorrect index behaviour with rtree on box values
Date: Mon, 24 Jan 2005 23:29:12 -0000
Organization: http://www.supernews.com - all your nntp are belong to us
Message-ID: <slrncvb167.5vn.andrew+nonews@trinity.supernews.net>
Reply-To: andrew@supernews.com
User-Agent: slrn/0.9.8.0 (FreeBSD)
X-Complaints-To: abuse@supernews.com
Lines: 56
To: pgsql-bugs@postgresql.org
X-Virus-Scanned: by amavisd-new at hub.org
X-Mailing-List: pgsql-bugs
Precedence: bulk
Sender: pgsql-bugs-owner@postgresql.org
X-Virus-Scanned: by amavisd-new at hub.org
X-Spam-Checker-Version: SpamAssassin 2.61 (1.212.2.1-2003-12-09-exp) on 
	candle.pha.pa.us
X-Spam-Status: No, hits=-4.9 required=5.0 tests=BAYES_00 autolearn=ham 
	version=2.61
Status: OR

Testcase:

create table boxtest (a box);
create index boxtest_idx on boxtest using rtree (a);
 
create function gen_data() returns void as '
  begin for i in 1..200 loop
    insert into boxtest
     values (box(point((i*2-1)::float,0),point((i*2)::float,1)));
  end loop;
  return;
end;' language plpgsql;
 
select gen_data();
analyze boxtest;
 
set enable_seqscan = false;
set enable_bitmapscan = true;
set enable_indexscan = true;
select * from boxtest where a << '(3,0),(3,1)'::box;
set enable_seqscan = true;
set enable_bitmapscan = false;
set enable_indexscan = false;
select * from boxtest where a << '(3,0),(3,1)'::box;


Those two selects at the end should clearly return the same result, a
single row. In fact, what happens is that the second returns no rows at
all; I tested this on 7.4.6, but others have confirmed this on everything
from 7.3 to latest.

The problem is that the semantics of the &< and &> operators for the box
type are not what rtree needs for the "OverLeft" and "OverRight" slots of
the operator class. Specifically, what rtree needs is this:

  if X << K or X &< K
  then for all A where A is a union of values including X,
  then A &< K

(the designation "&<" is of course arbitrary, what matters is what operator
is placed in the applicable slot of the opclass. Same goes for >> and &>.)

This is because rtree converts (see rtstrat.c) the original "Left" operator
to an "OverLeft" when comparing against internal nodes of the index, which
contain values which are the union of all values in their subtree. In the
testcase, the top node of the tree contains as its first entry a union
value of the form (184,1),(1,0), which the scan then rejects since
(184,1),(1,0) &< (3,0),(3,1) is false.

I can see three possible approaches to fixing this:

1) change the semantics of &< and &> to match rtree's expectations

2) replace &< and &> in the opclass with operators that behave as rtree
expects (this will have the side effect of rendering &< and &> un-indexable)

3) change rtree's behaviour in some way.

-- 
Andrew, Supernews
http://www.supernews.com - individual and corporate NNTP services

---------------------------(end of broadcast)---------------------------
TIP 2: you can get off all lists at once with the unregister command
    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)

From pgsql-bugs-owner+M10748=pgman=candle.pha.pa.us@postgresql.org Mon Jan 24 18:57:46 2005
Return-path: <pgsql-bugs-owner+M10748=pgman=candle.pha.pa.us@postgresql.org>
Received: from svr1.postgresql.org (svr1.postgresql.org [200.46.204.71])
	by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id j0P0vjw18152
	for <pgman@candle.pha.pa.us>; Mon, 24 Jan 2005 19:57:45 -0500 (EST)
Received: from localhost (unknown [200.46.204.144])
	by svr1.postgresql.org (Postfix) with ESMTP id 1CD183A52F3
	for <pgman@candle.pha.pa.us>; Tue, 25 Jan 2005 00:57:41 +0000 (GMT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024)
	with ESMTP id 43652-07 for <pgman@candle.pha.pa.us>;
	Tue, 25 Jan 2005 00:57:39 +0000 (GMT)
Received: from postgresql.org (svr1.postgresql.org [200.46.204.71])
	by svr1.postgresql.org (Postfix) with ESMTP id 9FEA63A52C5
	for <pgman@candle.pha.pa.us>; Tue, 25 Jan 2005 00:57:40 +0000 (GMT)
X-Original-To: pgsql-bugs-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.144])
	by svr1.postgresql.org (Postfix) with ESMTP id 3E2AF3A1A35
	for <pgsql-bugs-postgresql.org@localhost.postgresql.org>; Tue, 25 Jan 2005 00:09:47 +0000 (GMT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024)
	with ESMTP id 39621-06
	for <pgsql-bugs-postgresql.org@localhost.postgresql.org>;
	Tue, 25 Jan 2005 00:09:42 +0000 (GMT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id CAB643A19B8
	for <pgsql-bugs@postgresql.org>; Tue, 25 Jan 2005 00:09:41 +0000 (GMT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j0P09fcc027307;
	Mon, 24 Jan 2005 19:09:42 -0500 (EST)
To: andrew@supernews.com
cc: pgsql-bugs@postgresql.org
Subject: Re: [BUGS] incorrect index behaviour with rtree on box values 
In-Reply-To: <slrncvb167.5vn.andrew+nonews@trinity.supernews.net> 
References: <slrncvb167.5vn.andrew+nonews@trinity.supernews.net>
Comments: In-reply-to Andrew - Supernews <andrew+nonews@supernews.com>
	message dated "Mon, 24 Jan 2005 23:29:12 +0000"
Date: Mon, 24 Jan 2005 19:09:41 -0500
Message-ID: <27306.1106611781@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>
X-Virus-Scanned: by amavisd-new at hub.org
X-Mailing-List: pgsql-bugs
Precedence: bulk
Sender: pgsql-bugs-owner@postgresql.org
X-Virus-Scanned: by amavisd-new at hub.org
Status: OR

Andrew - Supernews <andrew+nonews@supernews.com> writes:
> The problem is that the semantics of the &< and &> operators for the box
> type are not what rtree needs for the "OverLeft" and "OverRight" slots of
> the operator class.

This was observed nearly a year ago, see this thread:
http://archives.postgresql.org/pgsql-general/2004-03/msg01135.php

but apparently no one cares enough to fix it.  Are you volunteering?

			regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend

From pgsql-bugs-owner+M10762=pgman=candle.pha.pa.us@postgresql.org Wed Jan 26 08:56:08 2005
Return-path: <pgsql-bugs-owner+M10762=pgman=candle.pha.pa.us@postgresql.org>
Received: from svr1.postgresql.org (svr1.postgresql.org [200.46.204.71])
	by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id j0QEu6w07027
	for <pgman@candle.pha.pa.us>; Wed, 26 Jan 2005 09:56:07 -0500 (EST)
Received: from localhost (unknown [200.46.204.144])
	by svr1.postgresql.org (Postfix) with ESMTP id 86BC83A5C12
	for <pgman@candle.pha.pa.us>; Wed, 26 Jan 2005 14:56:02 +0000 (GMT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024)
	with ESMTP id 26111-04 for <pgman@candle.pha.pa.us>;
	Wed, 26 Jan 2005 14:55:58 +0000 (GMT)
Received: from postgresql.org (svr1.postgresql.org [200.46.204.71])
	by svr1.postgresql.org (Postfix) with ESMTP id 5329B3A5C0B
	for <pgman@candle.pha.pa.us>; Wed, 26 Jan 2005 14:56:02 +0000 (GMT)
X-Original-To: pgsql-bugs-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.144])
	by svr1.postgresql.org (Postfix) with ESMTP id 3C43C3A5801
	for <pgsql-bugs-postgresql.org@localhost.postgresql.org>; Wed, 26 Jan 2005 14:54:51 +0000 (GMT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024)
	with ESMTP id 25627-10
	for <pgsql-bugs-postgresql.org@localhost.postgresql.org>;
	Wed, 26 Jan 2005 14:54:39 +0000 (GMT)
Received: from floppy.pyrenet.fr (floppy.pyrenet.fr [194.250.190.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 17AD33A516B
	for <pgsql-bugs@postgresql.org>; Wed, 26 Jan 2005 14:54:42 +0000 (GMT)
Received: by floppy.pyrenet.fr (Postfix, from userid 106)
	id F3BF931D93; Wed, 26 Jan 2005 15:54:43 +0100 (MET)
From: Andrew - Supernews <andrew+nonews@supernews.com>
X-Newsgroups: pgsql.bugs
Subject: Re: [BUGS] incorrect index behaviour with rtree on box values
Date: Wed, 26 Jan 2005 14:54:41 -0000
Organization: http://www.supernews.com - all your nntp are belong to us
Message-ID: <slrncvfbph.5vn.andrew+nonews@trinity.supernews.net>
References: <slrncvb167.5vn.andrew+nonews@trinity.supernews.net> <27306.1106611781@sss.pgh.pa.us>
Reply-To: andrew@supernews.com
User-Agent: slrn/0.9.8.0 (FreeBSD)
X-Complaints-To: abuse@supernews.com
Lines: 79
To: pgsql-bugs@postgresql.org
X-Virus-Scanned: by amavisd-new at hub.org
X-Mailing-List: pgsql-bugs
Precedence: bulk
Sender: pgsql-bugs-owner@postgresql.org
X-Virus-Scanned: by amavisd-new at hub.org
X-Spam-Checker-Version: SpamAssassin 2.61 (1.212.2.1-2003-12-09-exp) on 
	candle.pha.pa.us
X-Spam-Status: No, hits=-4.9 required=5.0 tests=BAYES_00 autolearn=ham 
	version=2.61
Status: OR

On 2005-01-25, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Andrew - Supernews <andrew+nonews@supernews.com> writes:
>> The problem is that the semantics of the &< and &> operators for the box
>> type are not what rtree needs for the "OverLeft" and "OverRight" slots of
>> the operator class.
>
> This was observed nearly a year ago, see this thread:
> http://archives.postgresql.org/pgsql-general/2004-03/msg01135.php
>
> but apparently no one cares enough to fix it.  Are you volunteering?

Possibly. I don't feel comfortable with changing anything specific to the
geometric operators, since (a) I don't actually use them (I discovered
this issue when adding rtree support to a type of my own) and (b) the
compatibility implications are obvious. But I think there is a solution
that involves only changes to the rtree strategy code.

Looking at the earlier discussion: it seems to have ended with the
conclusion that &< should mean "does not extend to the right of", which
matches the current implementation for box, but not for some other types.

So for box values, we seem (and someone please correct me if I'm wrong) to
have the following semantics:

a << b   - a is strictly left of b, i.e. a.right < b.left
a &< b   - a is no further right than b, i.e. a.right <= b.right
a &> b   - a is no further left than b, i.e. a.left >= b.left
a >> b   - a is strictly right of b, i.e. a.left > b.right

For rtree to work as apparently intended, it needs four more operators,
to use for inner nodes when the scan operator is one of the above four.
However, a small modification to the way that the internal scan key is
initialised should eliminate the requirement to explicitly specify these
operators, which strikes me as the solution which preserves maximum
compatibility.  The four operators required are:

  NOT (a &> b)      (used when the scan operator is (a << b))
  NOT (a >> b)      (used when the scan operator is (a &< b))
  NOT (a << b)      (used when the scan operator is (a &> b))
  NOT (a &< b)      (used when the scan operator is (a >> b))

(This won't fix rtree on contrib/seg or contrib/cube, but those appear to be
broken already since they have different, and equally incorrect, definitions
of &> and &<. Fixing those would require slightly more complex operators,
such as NOT (a &> b OR a >> b) and so on. The more complex operators would
work for box too, so it might be worth using them anyway, but I don't yet
understand the scan key handling well enough to know if these can be
constructed rather than supplied in the opclass.)

Proof:

Let V be the scan key, i.e. the value we are searching for in the index.
Let U be a union over a set of values.
Let X be some value for which X OP V holds.

Consider an internal node entry with union U. We require that the following
holds: if U contains some value X where X OP V holds, then U OP' V must be
true. (But not the converse; U OP' V may be true even if no such X exists in
U. However, we wish it to be false as much as possible for efficiency.)

When OP is << :

X << V, therefore X.right < V.left, therefore X.left < V.left
therefore NOT (X &> V)

If U contains X, then U &> V is true iff U.left >= V.left

U.left <= min(E.left) for all elements E of U, and therefore for X if X in U

So if X in U, then U.left <= X.left < V.left, and therefore NOT (U &> V)

When OP is &< :

X &< V, therefore X.right <= V.right, therefore X.left <= V.right
therefore NOT (X >> V), and similar reasoning for U containing X as above.

-- 
Andrew, Supernews
http://www.supernews.com - individual and corporate NNTP services

---------------------------(end of broadcast)---------------------------
TIP 7: don't forget to increase your free space map settings