Date: Mon, 13 Jan 2020 21:12:38 +0100
From: Julien Rouhaud <rjuju123@gmail.com>
To: Tomas Vondra <tomas.vondra@2ndquadrant.com>
Cc: Konstantin Knizhnik <k.knizhnik@postgrespro.ru>,
	=?utf-8?B?5pu+5paH5peMKOS5ieS7jik=?= <wenjing.zwj@alibaba-inc.com>,
	Pavel Stehule <pavel.stehule@gmail.com>,
	Robert Haas <robertmhaas@gmail.com>,
	PostgreSQL Hackers <pgsql-hackers@postgresql.org>,
	=?utf-8?B?6JSh5p2+6ZyyKOWtkOWYiSk=?= <zijia@taobao.com>,
	"Cai, Le" <le.cai@alibaba-inc.com>,
	=?utf-8?B?6JCn5bCR6IGqKOmTgeW6tSk=?= <shaocong.xsc@alibaba-inc.com>
Subject: Re: [Proposal] Global temporary tables
Message-ID: <20200113201216.GA17684@nol>
References: <5767433E-1A77-4682-B3FD-4869911021C8@alibaba-inc.com>
 <6b7570aa-14ea-2f28-c3b5-74dd785bbafe@postgrespro.ru>
 <20200105200603.wrb4beacwb57hhpk@development>
 <33969545-BEF7-4518-A741-302CBF7FF269@alibaba-inc.com>
 <0498c45c-53cb-d9b0-c1e2-7d2d1a618511@postgrespro.ru>
 <20200109164829.mdl3z5xalmaqxs67@development>
 <3fef844f-0e73-8fcb-e4b6-595d6317b191@postgrespro.ru>
 <20200112015109.caab6qdr2ob6d5jv@development>
 <c5f9b7ab-9156-f2f4-d162-99448c3aecfc@postgrespro.ru>
 <20200113163253.ja34iqsors3o44sv@development>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200113163253.ja34iqsors3o44sv@development>
Precedence: bulk

On Mon, Jan 13, 2020 at 05:32:53PM +0100, Tomas Vondra wrote:
> On Mon, Jan 13, 2020 at 11:08:40AM +0300, Konstantin Knizhnik wrote:
> >
> >"if any code tried to access the statistics directly from the table, 
> >rather than via the caches".
> >
> >Currently optimizer is accessing statistic though caches. So this 
> >approach works. If somebody will rewrite optimizer or provide own 
> >custom optimizer in extension which access statistic directly
> >then it we really be a problem. But I wonder why bypassing catalog 
> >cache may be needed.
> >
> 
> I don't know, but it seems extensions like hypopg do it.

AFAIR, hypopg only opens pg_statistic to use its tupledesc when creating
statistics on hypothetical partitions, but it should otherwise never reads or
need plain pg_statistic rows.