Re: Indices for select count(*)?

>
> I wouldn't mind a "with visibility" switch for indexes that you could
> throw when creating them for this purpose. But burdening all indexes
> with this overhead when most wouldn't need it is not, IMHO, a good idea.
>

that would add complexity to the index code for... just one case?

what about a set of functions instead...

one function to create all necesary triggers to maintain a different
table with a count for the table, and one function that retrieves that
info

select start_counter_on_table('table_name');
select get_counter_on_table('table_name');

of course, this could be usefull just for the case of "select * from
table"... but that case is the whole problem...

--
regards,
Jaime Casanova
(DBA: DataBase Aniquilator ;)