Re: Success: Finished porting application to postgreSQL

On Mon, 19 Aug 2002, Ralph Graulich wrote:

> This is where the partitioned indexes - how I call them - come into play:

The correct term is "partial indexes."

And yeah, I think this is one of the most powerful features of postgres,
and one that should receive more press. I've been experimenting with
them a bit in a large application that I've been workig on (that Tsutaya
billion-row thing) and they also appear to be useful when you've got to
import a lot of data into an existing table, and don't have time to do a
full index rebuild.

I just wish that the query analyzer were a bit smarter about using
them. I.e., if I have an index on values 0-500M, and another on values
500M-600M, it would be nice if, when I limited my query to "value >= 0
and value <= 600M" it would use those two indexes for the appropriate
parts of the table.

> * I want to look into defining my own datatypes and objects, which I think
> can make things for me even more easy.

This can be really handy, too. In another application I'm working
on, I have a table of (classless) networks, and need to find out
which one best matches an arbitrary IP address. Dead simple and
pretty efficient in postgres, becuase the type to do this is there.

Anyway, this whole testimonial of yours should get put up somewhere
on our web site. It's really valuable stuff.

cjs
--
Curt Sampson  <cjs@cynic.net>   +81 90 7737 2974   http://www.netbsd.org
    Don't you know, in this new Dark Age, we're all light.  --XTC