On 12/15/11, Marti Raudsepp <marti@juffo.org> wrote:
> On Thu, Dec 15, 2011 at 18:10, Robert James <srobertjames@gmail.com> wrote:
>> How do I do the equivalent of an ANY() or ALL() in PG Aggregate SQL?
>
> Note that in many cases, writing an EXISTS(SELECT ...) or NOT
> EXISTS(...) subquery is faster, since the planner can often optimize
> those to a single index access -- whereas an aggregate function would
> necessarily need to walk through and evaluate all potential rows.
>
Really? The planner can't tell that, for instance, BOOL_AND (false, *)
is automatically false?
(BTW Thanks for the great responses everyone! On this and other posts
on this list)