Re: Ugly group by problem
От | Achilleus Mantzios |
---|---|
Тема | Re: Ugly group by problem |
Дата | |
Msg-id | Pine.LNX.4.44.0603301317350.23284-100000@matrix.gatewaynet.com обсуждение исходный текст |
Ответ на | Re: Ugly group by problem (Markus Schaber <schabi@logix-tt.com>) |
Список | pgsql-sql |
O Markus Schaber έγραψε στις Mar 29, 2006 : > Hi, Achilleus, > > Achilleus Mantzios wrote: > > > foodb=# SELECT qoo.foo2,sum(qoo.foo3) from (SELECT mt.link_id as > > foo,_int_union(array(select mt2.feat_id from markustest mt2 where > > mt2.link_id=mt.link_id order by mt2.feat_id),'{}') as foo2,other::int4 as > > foo3 from markustest mt) as qoo GROUP BY qoo.foo2; > > foo2 | sum > > --------+----- > > {2} | 1 > > {5,23} | 13 > > {23} | 14 > > (3 rows) > > This is much like I intended to do it, but using "select distinct" in > the inner select as I don't have _int_union here, and using a temporary > table to collect the sets of link ids. Just some facts: sort(uniq(int[])) has the same effect as select DISTINCT... ORDER BY DISTINCT. ORDER BY is ofcourse more intuitive and does not need contrib/intarray. In fact , i dont even remember why i didnt write the query with DISTINCT... in the first place :) > > Markus > > > -- -Achilleus
В списке pgsql-sql по дате отправления: