Message-ID: <c886cc0645ac17f6fe3d5b155e01d1be22bd6c97.camel@cybertec.at>
Subject: Re: BUG #19740: `has_language_privilege` returns TRUE for a
 nonexistent language OID when the user is a superuser
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: theshallow27@gmail.com, pgsql-bugs@lists.postgresql.org
Date: Sun, 04 Oct 2026 15:39:32 +0200
In-Reply-To: <19740-677922f8a0cc921c@postgresql.org>
References: <19740-677922f8a0cc921c@postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.60.2 (3.60.2-2.fc44) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/c886cc0645ac17f6fe3d5b155e01d1be22bd6c97.camel%40cybertec.at>
Precedence: bulk

On Fri, 2026-10-02 at 22:38 +0000, PG Bug reporting form wrote:
> ostgreSQL version: 18.6
>=20
> For a nonexistent language OID, the implicit-current-user form returns TR=
UE
> when the current user is a superuser. An explicit non-superuser role retu=
rns
> NULL for the same missing OID. This makes the result depend on the role's
> superuser status even though the referenced language does not exist.
>=20
> **Reproduction:** Run as the default `postgres` superuser:
>=20
> ```sql
> SELECT NOT EXISTS (SELECT FROM pg_language WHERE oid =3D 0),
>        has_language_privilege(0::oid, 'USAGE'),
>        has_language_privilege('pg_monitor', 0::oid, 'USAGE');
> ```
>=20
> **Actual result:** `true | true | NULL`.
>=20
> **Expected result:** The privilege inquiry should return NULL for the
> missing
> language OID in both forms, matching the function's missing-object handli=
ng
> for non-superuser roles.

I agree that that is not so great.  Since this behavior is in the function
object_aclcheck_ext(), which is used by all the has_*_privilege functions,
the same oddity affects all those functions.

I think that would be easy to change, but I wonder if it is a good idea.
That bug hardly hurts: has_language_privilege() is typically not what you
use to find out if a procedural language exists or not.
And perhaps there is a misguided script somewhere out there that would
suddenly start to misbehave if a superuser no longer is reported as having
all privileges on non-existing objects.

Is it a real problem for you?

Yours,
Laurenz Albe