Re: "grant usage on schema" confers the ability to execute all user-defined functions in that schema, with needing to grant "execute"

On Fri, Feb 11, 2022 at 05:05:20PM -0500, Tom Lane wrote:
> Bryn Llewellyn <bryn@yugabyte.com> writes:
> > I confess that I'm surprised by the choice of the default behavior. It seems to be at odds with the principle of
leastprivilege that insists that  you actively opt in to any relevant privilege.
 
> 
> I'd be the first to agree that this behavior sacrifices security
> principles for convenience.  However, it's not that big a deal
> in practice, because functions that aren't SECURITY DEFINER can't
> do anything that the caller couldn't do anyway.  You do need to
> be careful about the default PUBLIC grant if you're making a
> SECURITY DEFINER function, but that's a minority use-case.

How would you do that securely?  Create the function and set its
permissions in a transaction block?

> (I wonder if it'd be practical or useful to emit a warning when
> granting permissions on an object that already has a grant of
> the same permissions to PUBLIC.  That would at least cue people
> who don't understand about this behavior that they ought to look
> more closely.)

Agreed.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  If only the physical world exists, free will is an illusion.