Andreas Lehrbaum <andreas(dot)lehrbaum(at)kabelweb(dot)at> writes:
> Is it possible in _any_ way to control the order of the rows passed to an
> aggregate function?
For simple aggregation you can do
SELECT myagg(col) FROM (SELECT col FROM ... ORDER BY foo) ss;
If you wanted to group, you could try
SELECT gcol, myagg(col) FROM
(SELECT gcol, col FROM ... ORDER BY gcol, foo) ss
GROUP BY gcol;
but this does *not* work reliably in current releases (it will work in
7.4 though). See the archives.
regards, tom lane