Robert,
* Robert Haas (robertmhaas@gmail.com) wrote:
> On Wed, Nov 9, 2016 at 2:54 PM, Corey Huinker <corey.huinker@gmail.com> wrote:
> > 3. The operation is judged to have succeeded if at least one permission is
> > granted, or NO grants failed (i.e. there was nothing to grant).
>
> Allow me to be skeptical. If a user types INSERT INTO blah VALUES
> (...), (...), (...) should we change the system to report success if
> at least 1 of the 3 rows got successfully inserted? I bet that
> wouldn't go over well.
To this point, we already do this for GRANT and REVOKE, so if this is
going to be based around those commands then it should perform in a
similar manner. Of course, that behavior is required for SQL spec and
as Tom points out that might be reason enough to avoid actually tying
this in with GRANT/REVOKE since we could end up in a tough spot if the
SQL committee decides to take a different direction than what we use (or
use the keywords we pick for something else).
Thanks!
Stephen