Re: psql: Add role's membership options to the \du+ command - Mailing list pgsql-hackers

From Pavel Luzanov
Subject Re: psql: Add role's membership options to the \du+ command
Date
Msg-id 94ce7392-fa90-bc1c-4dc2-677161bbff58@postgrespro.ru
Whole thread Raw
In response to Re: psql: Add role's membership options to the \du+ command  ("David G. Johnston" <david.g.johnston@gmail.com>)
Responses Re: psql: Add role's membership options to the \du+ command
List pgsql-hackers
On 17.02.2023 19:53, David G. Johnston wrote:
On Fri, Feb 17, 2023 at 4:02 AM Pavel Luzanov <p.luzanov@postgrespro.ru> wrote:
                                   List of roles
 Role name |                         Attributes                         | Member of
-----------+------------------------------------------------------------+-----------
 admin     | Create role                                                | {bob,bob}
 bob       |                                                            | {}
 postgres  | Superuser, Create role, Create DB, Replication, Bypass RLS | {}

First 'grant bob to admin' command issued immediately after creating role bob by superuser(grantor=10). Second command issues by admin role and set membership options SET and INHERIT.If we don't ready to display membership options with \du+ may be at least we must group records in 'Member of' column for \du command?


I agree that these views should GROUP BY roleid and use bool_or(*_option) to produce their result. 

Ok, I'll try in the next few days. But what presentation format to use?

1. bob(admin_option=t inherit_option=t set_option=f) -- it seems very long
2. bob(ai) -- short, but will it be clear?
3. something else?

Their purpose is to communicate the current effective state to the user, not facilitate full inspection of the configuration, possibly to aid in issuing GRANT and REVOKE commands.

This can help in issuing GRANT command, but not REVOKE. Revoking a role's membership is now very similar to revoking privileges. Only the role that granted membership can revoke that membership. So for REVOKE you need to know who granted membership, but this information will not be available after grouping.

One thing I found, and I plan to bring this up independently once I've collected my thoughts, is that pg_has_role() uses the terminology "USAGE" and "MEMBER" for "INHERIT" and "SET" respectively.

It's annoying that "member" has been overloaded here.  And the choice of USAGE just seems arbitrary (though I haven't researched it) given the related syntax.



I didn't even know this function existed. But I see that it was changed in 3d14e171 with updated documentation:
https://www.postgresql.org/docs/devel/functions-info.html#FUNCTIONS-INFO-ACCESS
Maybe that's enough.

-- 
Pavel Luzanov
Postgres Professional: https://postgrespro.com

pgsql-hackers by date:

Previous
From: Andres Freund
Date:
Subject: Re: refactoring relation extension and BufferAlloc(), faster COPY
Next
From: Peter Eisentraut
Date:
Subject: pg_dump: Remove some dead code