On Mon, Apr 10, 2017 at 12:01 PM, Tom Lane <tgl@xxxxxxxxxxxxx> wrote: > Merlin Moncure <mmoncure@xxxxxxxxx> writes: >> On Sun, Apr 9, 2017 at 4:27 PM, Guyren Howe <guyren@xxxxxxxxx> wrote: >>> If I do a SELECT with a GROUP_BY and multiple ARRAY_AGGs, will the ARRAY_AGGs be guaranteed to have entries in the same (ie corresponding) order? >>> >>> eg >>> >>> SELECT >>> u.name, >>> ARRAY_AGG(o.order_date) AS order_dates, >>> ARRAY_AGG(o.order_total) AS order_totals >>> FROM >>> user u JOIN >>> orders o USING (user_id) >>> GROUP BY >>> u.user_id > >> It is unsafe to rely on aggregation order unless specified -- you can >> add ORDER BY to the aggregation clause. > > You definitely can't assume anything about the order in which the FROM > clause will deliver rows, but I think that's not quite what the question > was. If I read it right, the OP wants to be sure that the two aggregate > functions will see the data in the *same* unspecified order. I think > that's a pretty safe assumption. The server would have to go way > out of its way to do differently, and it doesn't. Sure, but isn't it fair to consider that an implementation artifact? If his code depends on that ordering being the same across aggregate functions, and the SQL standard doesn't specify that (I guess it might, but I'm skeptical), he ought to specify that for clarify at the very least. merlin -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
