From: Tom Lane <tgl@sss.pgh.pa.us>
To: "Ghiurea, Isabella" <Isabella.Ghiurea@nrc-cnrc.gc.ca>
cc: "pgsql-admin@postgresql.org" <pgsql-admin@postgresql.org>
Subject: Re: PGSQL 10.9 vs PGSQL 9.6 SQL query behaviour.
In-reply-to: <1571677730032.95714@nrc-cnrc.gc.ca>
References: <1571677730032.95714@nrc-cnrc.gc.ca>
Comments: In-reply-to "Ghiurea, Isabella" <Isabella.Ghiurea@nrc-cnrc.gc.ca>
	message dated "Mon, 21 Oct 2019 17:08:50 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <4111.1571685493.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Mon, 21 Oct 2019 15:18:13 -0400
Message-ID: <4112.1571685493@sss.pgh.pa.us>
Precedence: bulk

"Ghiurea, Isabella" <Isabella.Ghiurea@nrc-cnrc.gc.ca> writes:
> we are facing some  performance issues with same SQL running in differen=
t version  ( PG SQL 10.9 vs
> 9.5.16)returning  different results when using GIST spatial index  see b=
ellow ,

Getting the wrong answer isn't a "performance issue" :-(

You might find that reindexing the index makes the problem go away.
Otherwise, I'd suggest inquiring in the PostGIS mailing list whether
there's any known bugs in their code for 10.x.  I don't see anything
in our commit logs suggesting that there's a known problem with the
core GIST code, but maybe there's something wrong with the PostGIS
operator class the index is using.

			regards, tom lane