Message-ID: <5184088806e70ed5db7e83bd6540a016b3802c1d.camel@cybertec.at>
Subject: Re: BUG #19483: pg_upgrade fails with orphan records in
 pg_init_priv catalog table
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: =?ISO-8859-1?Q?H=FCseyin?= Demir <huseyin.d3r@gmail.com>, Tom Lane
	 <tgl@sss.pgh.pa.us>
Cc: Greg Sabino Mullane <htamfids@gmail.com>, pgsql-bugs@lists.postgresql.org
Date: Sun, 21 Jun 2026 07:47:33 +0200
In-Reply-To: <d305af372949be0c2c4155168ca7ba97e160bc52.camel@cybertec.at>
References: <19483-80de42dc4e62cfd6@postgresql.org>
		 <CAKAnmmLYXcBSV8i5MVuiwu8tkX-JDGmZtJ38_YbdUTwoDW7xKg@mail.gmail.com>
		 <104812.1780847519@sss.pgh.pa.us>
		 <CAB5wL7bo-7okBCJ24fdVn4UhnuPLPG22eQjaLJZ9GJtv0FEWMg@mail.gmail.com>
		 <CAB5wL7bunPCi93Bwi7zN_1gFAZhKLwS=Fsub91SW417xujGtNA@mail.gmail.com>
		 <CAB5wL7aKGFFEF1SM54GxvaQPH=7Bs=HuUCqi7F4Ve025+Z950A@mail.gmail.com>
		 <CAB5wL7aK6UaukMSegzEG68y0VrKOg7wFnPNRaDaHreBf2iVWag@mail.gmail.com>
	 <d305af372949be0c2c4155168ca7ba97e160bc52.camel@cybertec.at>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.58.3 (3.58.3-1.fc43) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/5184088806e70ed5db7e83bd6540a016b3802c1d.camel%40cybertec.at>
Precedence: bulk

On Sat, 2026-06-20 at 23:53 +0200, I wrote:
> I think that we need something like this fix, because a failing
> upgrade is a bug.  For the same reason I think that the fix
> should be backpatched.
>=20
> I looked at your patch and found that the query you added doesn't
> cover the important case where the grantor is a non-existing role
>=20
> I suggest a query like this one: [...]

Further testing shows that changing this query isn't enough.
There are three more places where pg_dump queries pg_init_privs
(in getAggregates, getFuncs and dumpTable).

So we'd have to use a similarly ugly query in all these places,
which doesn't seem particularly attractive and introduces
considerable code duplication.

One approach I can think of is to have a macro SAFE_INITPRIVS
that contains the ugly subquery and is used in all these places.

The other idea is to do some post-processing of the aclitems
found, but they are in string form and would need to get parsed
again, which doesn't look attractive either.

Yours,
Laurenz Albe