Message-ID: <aaaeed9d3c93868cf55c0e6279f6a48189b12988.camel@j-davis.com>
Subject: Re: [17] Special search_path names "!pg_temp" and "!pg_catalog"
From: Jeff Davis <pgsql@j-davis.com>
To: Pavel Stehule <pavel.stehule@gmail.com>
Cc: pgsql-hackers@postgresql.org
Date: Mon, 21 Aug 2023 09:08:43 -0700
In-Reply-To: <CAFj8pRCT20Zg-H3+QYuDvNNGMHyAMtHkvqzUgJgJxx7FwPhz6w@mail.gmail.com>
References: <a6865db287596c9c6ea12bdd9de87216cb5e7902.camel@j-davis.com>
	 <CAFj8pRCT20Zg-H3+QYuDvNNGMHyAMtHkvqzUgJgJxx7FwPhz6w@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4-0ubuntu2 
MIME-Version: 1.0
Archived-At: <https://www.postgresql.org/message-id/aaaeed9d3c93868cf55c0e6279f6a48189b12988.camel%40j-davis.com>
Precedence: bulk

On Sat, 2023-08-19 at 07:18 +0200, Pavel Stehule wrote:
> cannot be better special syntax
>=20
> CREATE OR REPLACE FUNCTION xxx()
> RETURNS yyy AS $$ ... $$$
> SET SEARCH_PATH DISABLE
>=20
> with possible next modification
>=20
> SET SEARCH_PATH CATALOG .. only for pg_catalog
> SET SEARCH_PATH MINIMAL .. pg_catalog, pg_temp

I agree that we should consider new syntax, and there's a related
discussion here:

https://www.postgresql.org/message-id/flat/2710f56add351a1ed553efb677408e51=
b060e67c.camel@j-davis.com

Regardless, even with syntax changes, we need something to print when
someone does a "SHOW search_path", i.e. some representation that
indicates pg_temp is excluded. That way it can also be saved and
restored.

> I question if we should block search path settings when this setting
> is used. Although I set search_path, the search_path can be
> overwritten in function of inside some nesting calls=20

If so, that should be a separate feature. For the purposes of this
thread, we just need a way to represent a search path that excludes
pg_temp and/or pg_catalog.

Regards,
	Jeff Davis