Hoang-Vu PHUNG schrieb:

> I'm still confused why select count(*) in firebird is
> too slow compared to other databases. Of course, your
> trick can help but imagine if you should do for every
> big tables... Again, request the size of a file is
> quite fast. Finally, scan a file of 200'000 rows to
> get the number of rows is also fast done, even in perl
> / java ...

Another question is: what do you need a count(*) for at all?
(and why for every table???)
Retrieving the accurate number of records doesn't look like a really
useful information to me.

You know:
"The only statistics you can trust are those you falsified yourself".

Luc.