At 09:35 AM 20/10/2005 +0000, you wrote:

> > Back at you, to ask yourself why select count(*) is a "weakness" at
> > all? What on earth do you want it for? It has its possible uses
> > in aggregate queries, where it doesn't pose a performance problem,
> > but what is a sensible reason for counting all of the rows in a
> > table? And, if you do count them, which ones are you going to
> > count? and why?
>
>How about some OLAP-type queries:
>- how many orders did we ship last month?
>- how many customers have ordered widget X?
>Or think about a web forum, displaying the number of reads, comments,
>etc. for a specific article.

Er-hum, none of those examples does a count(*) on a whole table.

>I agree that index-based count(*) is probably not a must-have feature,
>but to discount it wholesale is a bit steep, I think.

Not discounting it at all. What I'm discounting is this obsession with
benchmarking the performance of databases on select count(*) from aTable,
as though it meant something.

Helen