On Sun, 2023-09-17 at 21:31 +0200, Erik Wienhold wrote:
> I wrote a patch to change psql's display of zero privileges after a user's
> reported confusion with the psql output for zero vs. default privileges [1].
> Admittedly, zero privileges is a rare use case [2] but I think psql should not
> confuse the user in the off chance that this happens.
>
> With this change psql now prints "(none)" for zero privileges instead of
> nothing. This affects the following meta commands:
>
> \db+ \dD+ \df+ \dL+ \dl+ \dn+ \dp \dT+ \l
>
> Default privileges start as NULL::aclitem[] in various catalog columns but
> revoking the default privileges leaves an empty aclitem array. Using
> \pset null '(null)' as a workaround to spot default privileges does not work
> because the meta commands ignore this setting.
>
> The privileges shown by \dconfig+ and \ddp as well as the column privileges
> shown by \dp are not affected by this change because those privileges are reset
> to NULL instead of leaving empty arrays.
>
> Commands \des+ and \dew+ are not covered in src/test/regress because no foreign
> data wrapper is available at this point to create a foreign server.
>
> [1] https://www.postgresql.org/message-id/efdd465d-a795-6188-7f71-7cdb4b2be031%40mtneva.com
> [2] https://www.postgresql.org/message-id/31246.1693337238%40sss.pgh.pa.us
Reading that thread, I had the impression that there was more support for
honoring "\pset null" rather than unconditionally displaying "(none)".
The simple attached patch does it like that. What do you think?
Yours,
Laurenz Albe