Re: count(*) and bad design was: Experiences with extensibility

On Wed, 09 Jan 2008 20:29:39 +0100
Zoltan Boszormenyi <zb@cybertec.at> wrote:

> The decision to use MVCC in PostgreSQL makes the point moot.

...

thanks.

> In PostgreSQL, COUNT(*) responds closely at the same speed
> regardless of other transactions. Which way do you prefer?

Considering the relative value of count my interest was for something
that is even less precise than the "usual" count but performs better.
I'm not proposing to turn Postgres into MySQL.

--
Ivan Sergio Borgonovo
http://www.webthatworks.it