On 2019-12-05 00:16, Tom Lane wrote:
> Seems reasonable. The argument for making this an exception to
> allow_system_table_mods was always more about expediency than logical
> cleanliness. After the recent changes I think it's okay to remove the
> special case (especially since nobody has griped about it being broken).
>
> However ... if we're not going to have that special case, couldn't
> we get rid of the whole business of having a special permissions test?
> What is it that ATSimplePermissions can't handle here? The business
> about requiring a colName certainly doesn't need to be done before the
> ownership check (in fact, it could be left to execution, so we don't need
> a prep function at all anymore).
Good point. Done in the attached patch.
(If someone wanted to revive the original functionality, it would
nowadays probably be easier to add a flag ATT_SYSTEM_TABLE to
ATSimplePermissions(), so there is really no reason to keep the old
function separate.)
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services