"Alloni Kramer" <alloni@aposiopesis.net> writes:
> Simplified version of problem: We have a table. We'll say it has four
> variables, a, b, c, and squelch, all ints. I want to get all values of
> these variables for which:
> c is within a certain range of values (call it between 3 and 5).
> for any given value of a, c is maximum.
> - and -
> for any given value of a for which c is maximal, b is maximum. That is,
> if we have the values of
This isn't perfectly clear, but I think what you are after can be done
with
select distinct on (a) * from mytab
where c between 3 and 5
order by a, c desc, b desc;
The "weather report" example in the SELECT reference page may be
helpful.
It's possible to do this without the nonstandard DISTINCT ON clause,
but it's a lot more tedious (and I forget just how at the moment ;-))
regards, tom lane