On Wed, Sep 24, 2014 at 1:37 PM, Alberto Cabello Sánchez <albe...@unex.es> wrote: > > On Wed, 24 Sep 2014 09:04:21 -0700 > Daniel Lenski <dlen...@gmail.com> wrote: > > > If I include the primary key of a table in my GROUP BY clause, PG 9.3 > > allows me to refer to other columns of that table without explicit GROUP BY: > > > > Why doesn't the same thing work with a non-NULL unique constraint? > > At first sight, primary key means no grouping at all, as there are no > duplicated A.primary_key values: > > SELECT A.document > FROM A > GROUP BY A.primary_key > > is the same as > > SELECT A.document > FROM A
I believe this blog post contains better examples of the feature he's referring to: http://www.depesz.com/2010/08/08/waiting-for-9-1-recognize-functional-dependency-on-primary-keys/ For example: SELECT p.id, p.firstname, p.lastname, count(*) FROM people p JOIN visits v on p.id = v.person_id GROUP BY p.id; Geoff
