Date: Thu, 9 Apr 2015 23:39:26 +0200 (CEST)
From: Andreas Joseph Krogh <andreas@visena.com>
To: "pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
Message-ID: <VisenaEmail.63.beee75c83c50054c.14ca0154fe9@tc7-visena>
Subject: Cannot get query to use btree-gin index when ORDER BY
MIME-Version: 1.0
Content-Type: multipart/related;
 boundary="----=_Part_138_187269198.1428615566683"
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

------=_Part_137_1839181217.1428615566683
Content-Type: multipart/related; 
	boundary="----=_Part_138_187269198.1428615566683"

------=_Part_138_187269198.1428615566683
Content-Type: multipart/alternative;
 boundary="----=_Part_139_569086253.1428615566698"

------=_Part_139_569086253.1428615566698
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Hi all. =C2=A0 Using PG-9.4 =C2=A0 I have a query which goes something like=
 this =C2=A0 The=20
simplified schema is like this: =C2=A0 CREATE TABLE message( id serial PRIM=
ARY KEY,=20
subject varchar NOT NULL, folder_id integer NOT NULL REFERENCES folder(id),=
=20
received timestamp not null, fts_all tsvector ); =C2=A0 create index=20
message_fts_all_folder_idx ON message using gin (fts_all, folder_id); =C2=
=A0 SELECT=20
m.id, m.subject =C2=A0 FROM message m WHERE m.folder_id =3D 1 AND m.fts_all=
 @@=20
to_tsquery('simple', 'foo') ORDER BY received DESC LIMIT 10; =C2=A0 ... =C2=
=A0 On my=20
dataset it uses an index I have on (folder_id, received DESC), then filters=
 the=20
result, which is not optimal when searching in > 1million messages and the=
=20
result is large and I'm only after the first (newest) 10. =C2=A0 What I'd l=
ike is to=20
have an index like this: =C2=A0 create index message_fts_all_folder_idx ON =
message=20
using gin (fts_all, folder_id, received DESC); =C2=A0 but GIN doesn't allow=
 ASC/DESC=20
modifiers. =C2=A0 Any hints on how to optimize this? =C2=A0 Thanks. =C2=A0 =
-- Andreas Joseph=20
Krogh CTO / Partner - Visena AS Mobile: +47 909 56 963 andreas@visena.com=
=20
<mailto:andreas@visena.com> www.visena.com <https://www.visena.com> =20
<https://www.visena.com>
------=_Part_139_569086253.1428615566698
Content-Type: text/html;charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div>Hi all.</div>

<div>=C2=A0</div>

<div>Using PG-9.4</div>

<div>=C2=A0</div>

<div>I have a query which goes something like this</div>

<div>=C2=A0</div>

<div>
<div>The simplified schema is like this:</div>

<div>=C2=A0</div>

<div>CREATE TABLE message(</div>

<div>id serial PRIMARY KEY,</div>

<div>subject varchar NOT NULL,</div>

<div>folder_id integer NOT NULL REFERENCES folder(id),</div>

<div>received timestamp not null,</div>

<div>
<div>fts_all tsvector</div>
</div>

<div>);</div>

<div>=C2=A0</div>

<div>
<div>create index message_fts_all_folder_idx ON message using gin (fts_all,=
 folder_id);</div>

<div>=C2=A0</div>
</div>
</div>

<div>SELECT m.id, m.subject</div>

<div>=C2=A0 FROM message m</div>

<div>WHERE m.folder_id =3D 1 AND m.fts_all @@ to_tsquery('simple', 'foo')</=
div>

<div>ORDER BY received DESC LIMIT 10;</div>

<div>=C2=A0</div>

<div>...
<div>=C2=A0</div>

<div>On my dataset it uses an index I have on (folder_id, received DESC), t=
hen filters the result, which is not optimal when searching in &gt; 1millio=
n messages and the result is large and I'm only after the first (newest) 10=
.</div>

<div>=C2=A0</div>

<div>What I'd like is to have an index like this:</div>

<div>=C2=A0</div>

<div>create index message_fts_all_folder_idx ON message using gin (fts_all,=
 folder_id, received DESC);</div>

<div>=C2=A0</div>

<div>but GIN doesn't allow ASC/DESC modifiers.</div>

<div>=C2=A0</div>

<div>Any hints on how to optimize this?</div>

<div>=C2=A0</div>

<div>Thanks.</div>

<div>=C2=A0</div>

<div class=3D"origo-email-signature">
<div>--
<div><span style=3D"font-family:arial,helvetica,sans-serif;"><strong>Andrea=
s Joseph Krogh</strong></span></div>

<div><span style=3D"font-family:arial,helvetica,sans-serif;">CTO / Partner<=
/span> - Visena AS</div>

<div><span style=3D"font-family:arial,helvetica,sans-serif;">Mobile: +47 90=
9 56 963</span></div>

<div><a href=3D"mailto:andreas@visena.com">andreas@visena.com</a></div>

<div><a href=3D"https://www.visena.com">www.visena.com</a></div>

<div style=3D"padding-top: 5px"><a href=3D"https://www.visena.com"><img alt=
=3D"" src=3D"cid:part_F7451590003560SL5B3"></a></div>
</div>
</div>
</div>
------=_Part_139_569086253.1428615566698--

------=_Part_138_187269198.1428615566683
Content-Type: image/png
Content-Transfer-Encoding: base64
Content-Disposition: inline
Content-ID: <part_F7451590003560SL5B3>
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------=_Part_138_187269198.1428615566683--

------=_Part_137_1839181217.1428615566683--