Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From: Andrew Dunstan <andrew@dunslane.net>
Mime-Version: 1.0 (1.0)
Subject: Re: getting "shell command argument contains a newline or carriage return:" error with pg_dumpall when db name have new line in double quote
Date: Sun, 6 Apr 2025 12:10:06 -0400
Message-Id: <64EEE0C9-0461-4412-BADA-1C108E98BC1B@dunslane.net>
References: <202504061523.tnumfmmidwtz@alvherre.pgsql>
Cc: Nathan Bossart <nathandbossart@gmail.com>,
 Mahendra Singh Thalor <mahi6run@gmail.com>,
 Srinath Reddy <srinath2133@gmail.com>, pgsql-hackers@lists.postgresql.org
In-Reply-To: <202504061523.tnumfmmidwtz@alvherre.pgsql>
To: =?utf-8?Q?=C3=81lvaro_Herrera?= <alvherre@alvh.no-ip.org>
Archived-At: <https://www.postgresql.org/message-id/64EEE0C9-0461-4412-BADA-1C108E98BC1B%40dunslane.net>
Precedence: bulk


Sent from my iPhone

> On Apr 6, 2025, at 11:23=E2=80=AFAM, =C3=81lvaro Herrera <alvherre@alvh.no=
-ip.org> wrote:
>=20
> =EF=BB=BFOn 2025-Apr-06, Andrew Dunstan wrote:
>=20
>> On 2025-03-28 Fr 10:43 AM, Nathan Bossart wrote:
>=20
>>> Taking a step back, are we sure that 1) this is the right place to do th=
ese
>>> checks and 2) we shouldn't apply the same restrictions to all names?  I'=
m
>>> wondering if it would be better to add these checks to the grammar inste=
ad
>>> of trying to patch up all the various places they are used in the tree.
>>=20
>> Maybe. I don't think there is time for that for v18, so we'd have to defe=
r
>> this for now. I can live with that - it's been like this for a long time.=

>=20
> Grumble.  I'd rather introduce a partial restriction now only for names
> that affect the tools failing outright (pg_dumpall in particular), than
> do nothing for this release.  If we feel the need to extend the
> restriction later, that's easy to do and bothers nobody.  We've wanted
> these characters to be forbidden on database names for a long time, but
> nobody has cared as much for their use on other types of names.  I'm not
> even sure we'd support the idea of forbidding them on all names (though
> the SQL standard doesn't allow control chars in identifiers.)

Ok. That seems reasonable, let=E2=80=99s do this now and revisit in the v19 c=
ycle=20

Cheers

Andrew