Jeff Davis <[EMAIL PROTECTED]> writes:
> I had an idea about using aggregates: what if I made an aggregate function
> called "first" that just returned the value in the first tuple it
> encountered?
You could make that work in 7.4, but not in any existing releases.
The trouble is that you need something like
SELECT first(foo) FROM (SELECT ... ORDER BY col1,col2) ss
GROUP BY col1
and before 7.4 the optimizer doesn't realize that it can skip re-sorting
at the outer level. So unless the sort is stable (which it won't be, on
most platforms anyway) the needed ordering by col2 within each group is
destroyed.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
