Hello all,
I wanted to verify what we are seeing.
Select a, aggregate(b)
from c
group by a
order by a,b
Is not accepted by postgres. This will only work if you order by a. But,
this means that the records that are grouped are processed in no apparent
order.
We have some custom aggregate functions where the order of the rows is
important. Is there no way to do this without a subselect?
select a,aggregate(b)
from (
select a,b
from c
order by a,b
) as foo
group by a
order by a
Even with a subselect doing the ordering ahead of time, is there any
guarrantee that the records will be processed in the group by with the
specified order? Or will the group by always be arbitrary in it's ordering
of the records?
This is pg 8.1 on Linux.
Thanks,
-chris
--
Chris Kratz
