Date: Fri, 19 Jun 2026 13:36:45 +0000
From: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
To: Jeff Davis <pgsql@j-davis.com>
Cc: Heikki Linnakangas <hlinnaka@iki.fi>,
	Robert Haas <robertmhaas@gmail.com>,
	Roman Eskin <r.eskin@arenadata.io>,
	Michael Paquier <michael@paquier.xyz>,
	Alexander Lakhin <exclusion@gmail.com>,
	pgsql-hackers@lists.postgresql.org, Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re: Avoid orphaned objects dependencies, take 3
Message-ID: <ajVF7YWf+pxs4cOf@bdtpg>
References: <aie26jMSoEMrHLaI@bdtpg>
 <0fc145b9b5cf3f59207cf4ca60270448a2891c46.camel@j-davis.com>
 <ailZCCmS0bGlNBfe@bdtpg>
 <ail/6I6mcitovsUo@bdtpg>
 <a9eba27eeceebe751490951f0cf631906d4ffd75.camel@j-davis.com>
 <ajEg7PrJYWnmB6zk@bdtpg>
 <ac52a6aa6be9ccf5cf43c44d009b6e2aa7e2c93d.camel@j-davis.com>
 <ajI0Tz9dIJvLGHNY@bdtpg>
 <5591d661ea8189f5c057f6b48095f742e0d772f9.camel@j-davis.com>
 <b178a9b776897045b8cfe2745bb4f0b178a4ee55.camel@j-davis.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <b178a9b776897045b8cfe2745bb4f0b178a4ee55.camel@j-davis.com>
Archived-At: 
 <https://www.postgresql.org/message-id/ajVF7YWf%2Bpxs4cOf%40bdtpg>
Precedence: bulk

Hi,

On Thu, Jun 18, 2026 at 07:13:38PM -0700, Jeff Davis wrote:
> On Thu, 2026-06-18 at 16:21 -0700, Jeff Davis wrote:
> > IIUC, we cannot have false positives (tracking ACL checks that
> > wouldn't
> > have caused an abort) nor can we have false negatives (missing an ACL
> > check that could cause an abort).
> 
> Idea: what if we check for changes in ACLs on the object, rather than
> whether it passes the check or not?
> 
> Then, if track an ACL check that wouldn't actually cause a failure,
> then it still might be acceptable to throw an error if the ACL changes.
> Still some details to sort out, so this is just an idea.

Yeah, I think I do prefer this idea. As you say, that could cause an error even
if the ACL change does not REVOKE anything on this object (say the ACL change is
a GRANT), but that should be rare in practice and probably much simpler to reason
about that way.

But I don't think tracking ACL changes would be enough though. I think we would
also need to track ROLE changes.

So what about?

- Save a copy of the object's ACL and compare at recheck time: If not the same,
then error out.
- Save the ROLE membership and compare at recheck time. If not the same, then
error out.

That way we cover both parts: the object's ACL and the ROLE membership.

That's just a high level idea, I can move forward and try to implement it.

Thoughts?

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com