On Mon, Aug 25, 2014 at 07:44:06PM -0400, Tom Lane wrote:
> A larger point is that we could easily consider RESET as meaning
> "remove this option *if it's applied to this relation*", which would
> mean that resetting a nonexistent option shouldn't be an error.
> If we don't define the action that way, then should RESET foo, where
> foo is a valid option that's not been set on the particular table,
> be an error? If not, what's the argument for allowing that case
> and not this one? Do we need a RESET IF EXISTS to cover that?
>
> Please revert and return the patch for further work/discussion.
> We had consensus on a vague idea, not the details of this particular
> patch.
OK, that makes sense. Reverted.
--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ Everyone has their own god. +