>>>>> "David" == David Raymond <David.Raymond@tomtom.com> writes:
David> I checked for other places where there was weirdness going on
David> and there are more.
You could try something like,
select name, lname
from (select name, lag(name) over (order by name) as lname
from big_table) s
where name < lname;
That should show all cases where the sort order was inconsistent with
the < operator (which obviously should never happen).
However, there's not much more I can do to help with this, since I don't
use Windows myself and have no useful access to any Windows system. You
might try and cut down the data to the smallest set that shows
inconsistencies using the above; particularly relevant is whether the
problem only shows up for external merge sorts or whether it happens for
in-memory sorts too.
--
Andrew (irc:RhodiumToad)