Reuven M. Lerner escribió:
> Hi, Webb Sprague. You wrote:
> >... but I see two seq scans in your explain in a loop -- this is
> >probably not good. If you can find a way to rewrite the IN clause
> >(either de-normalizing through triggers to save whatever you need on
> >an insert and not have to deal with a set, or by using except in the
> >query, or someing else more creative)...
> I would normally agree that an IN clause is a place to worry -- except
> that I'm using IN on a very small table, with about 4-5 rows. That
> might indeed affect things, and I could certainly pull out these values
> into a Perl variable that I then interpolate literally into the SQL.
> However, I have to assume that this wouldn't affect things all that much.
Don't assume -- measure. I had a query which ran orders of magnitude
faster because I interpolated the constant list in the big query. The
table from which the interpolated values were being extracted had about
30 rows or so.
--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.
