Very Computer

That's alright sweetie pie.

Quote:> I might miss it again, but asking never hurts. So: isn't UNIQUE
> inside the COUNT illegal?

I've never been arrested for it Dude!

Quote:> I mean, the query translates to something like: "determine
> the groups by ord_id, then determine the sub-groups by ship_num, then count, then filter"

Not really. Here are the options:

select ord_id, count(*)

which groups by ord_id then counts the number of rows in each group.

select ord_id, count(ship_num)

which groups by ord_id then counts the number of ship_num's in each group. So this is the same as count(*).

select ord_id, count(unique ship_num)

which groups by ord_id then counts the number of unique ship_num's in each group. So if you have two of the same ship_num's in the same group, it will only get counted once.

Try these selects in any database:

select colno, count(*)
from syscolumns
group by 1
order by 1
;
select colno, count(collength)
from syscolumns
group by 1
order by 1
;
select colno, count(unique collength)
from syscolumns
group by 1
order by 1

Quote:> So probably a sub-query or "group by ord_id, ship_num" should do the trick.

Have another beer Dude, you have exhausted my brain cell now. Happy BDay!

AB
----------------------------------------------------------------
Get your free email from AltaVista at http://altavista.iname.com