Laurenz Albe <laurenz.albe@cybertec.at> writes:
> On Mon, 2023-10-16 at 19:05 -0700, David G. Johnston wrote:
>> Reading both threads I'm not seeing any specific rejection of the
>> solution that we simply represent empty privileges as "(none)".
> Thanks for that summary. I prefer my version (simply display NULLs
> as NULLs), but I am not wedded to it. I had the impression that Tom
> would prefer that too, but is woried if the incompatibility introduced
> would outweigh the small benefit (of either patch).
> So it is clear that we don't have a consensus.
FWIW, my druthers are to make the describe.c queries honor \pset null
(not only for privileges, but anywhere else they fail to) and do
nothing beyond that. I think that'll generally reduce the surprise
factor, while anything else we might opt to do will increase it.
If that fails to garner a consensus, my second choice would be to
do that plus translate empty-but-not-null ACLs to "(none)".
regards, tom lane