From: Tom Lane <tgl@sss.pgh.pa.us>
To: =?utf-8?Q?=C3=81lvaro?= Herrera <alvherre@alvh.no-ip.org>
cc: Pg Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: bogus error message for ALTER TABLE ALTER CONSTRAINT
In-reply-to: <202503041822.aobpqke3igvb@alvherre.pgsql>
References: <202503041822.aobpqke3igvb@alvherre.pgsql>
Comments: In-reply-to =?utf-8?Q?=C3=81lvaro?= Herrera
 <alvherre@alvh.no-ip.org>
	message dated "Tue, 04 Mar 2025 19:22:22 +0100"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <195581.1741125028.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Tue, 04 Mar 2025 16:50:28 -0500
Message-ID: <195582.1741125028@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/195582.1741125028%40sss.pgh.pa.us>
Precedence: bulk

=3D?utf-8?Q?=3DC3=3D81lvaro?=3D Herrera <alvherre@alvh.no-ip.org> writes:
> I just discovered that trying to set a foreign key as NO INHERIT in
> ALTER TABLE ALTER CONSTRAINT returns an absurd error message:
> create table pk (a int primary key);
> create table fk (a int references pk);
> alter table fk alter constraint fk_a_fkey deferrable, alter constraint f=
k_a_fkey no inherit;
> ERROR:  ALTER CONSTRAINT statement constraints cannot be marked NO INHER=
IT

> The explanation is that somebody misunderstood what must be given to
> processCASbits in 2013.  The intended message is:
>   ERROR:  FOREIGN KEY constraints cannot be marked NO INHERIT

Hmm.  I agree that "ALTER CONSTRAINT statement" is off the
mark here, but I'm not convinced that "FOREIGN KEY" is entirely
on-point either.  The grammar has no way of knowing what kind of
constraint is being targeted.  I do see that ATExecAlterConstraint
currently rejects every other kind of constraint, but do we need
to think of a more generic phrase?

			regards, tom lane