Re: Reverse Index ... how to ...

On Wed, 5 Apr 2006, Tom Lane wrote:

> "Marc G. Fournier" <scrappy@postgresql.org> writes:
>> I'm still searching through Google and whatnot, but not finding anything
>> off the bat ... is there some way of creating a 'REVERSE INDEX' on a
>> column in a table?
>
>> For instance, when I do a 'CLUSTER' to sort a table based on an INDEX, I'd
>> like to sort it in reverse order, so would need the INDEX to go from
>> 'GREATEST to LOWEST', vs 'LOWEST to GREATEST' ...
>
> You shouldn't need to worry about that during CLUSTER, as the system is
> perfectly capable of scanning an index in either forward or backward
> order at runtime.  For example,
>
> regression=# explain select * from tenk1 order by unique1;
>                                     QUERY PLAN
> ------------------------------------------------------------------------------------
> Index Scan using tenk1_unique1 on tenk1  (cost=0.00..1572.00 rows=10000 width=244)
> (1 row)
>
> regression=# explain select * from tenk1 order by unique1 desc;
>                                         QUERY PLAN
> ---------------------------------------------------------------------------------------------
> Index Scan Backward using tenk1_unique1 on tenk1  (cost=0.00..1572.00 rows=10000 width=244)
> (1 row)

Perfect, that was what I was looking for, thx ...

----
Marc G. Fournier           Hub.Org Networking Services (http://www.hub.org)
Email: scrappy@hub.org           Yahoo!: yscrappy              ICQ: 7615664