On Tue, Feb 11, 2003 at 09:48:11 -0500, Greg Stark <gsstark@mit.edu> wrote:
>
> So one of the items on the TODO list is "Add hash for evaluating GROUP BY
> aggregates (Tom)"
>
> I'm finding this would benefit a lot of my queries. Most of the time seems to
> be going into sorts for group by clauses. I don't know how long it would take
> to build a hash of course, but I suspect it would be less than the sort.
>
> Is this something a beginner could figure out? I'm thinking I need a normal
> Hash node that builds exactly the same kind of hash as a join, then a HashScan
> node that picks all the rows out of the hash.
This is already in 7.4. You could try it out by building from CVS.
From the HISTORY file:
System can use either hash- or sort-based strategy for grouped aggregation