> Well, date evidently isn't the high-order key of this index. But why
> exactly are you worried about a sort of 2 rows?
Aha that's nailed it:
usa=> explain select * from users_myfoods_map where user_id=1 and date
between '2003-11-03' and '2003-11-03' order by user_id, date; QUERY
PLAN
------------------------------------------------------------------------------------------------------------- Index
Scanusing users_myfoods_map_user_id_date_key on
users_myfoods_map (cost=0.00..3.78 rows=1 width=22) Index Cond: ((user_id = 1) AND (date >= '2003-11-03'::date) AND
(date <= '2003-11-03'::date))
(2 rows)
I don't care about this particular result. But imagine it running
thousands of times a minute, with result sets between 0 and 50 rows...
Chris
