At which point you're into a lot more complexity. If you're going to pre-generate that much data, just build a data set for each state that includes all zips within the greatest possible search radius, and start each search with the right data set.
-- Spring: Forces, Coiled Again!
Re: Re: Re: Re: Zipcode Proximity script
Replies are listed 'Best First'.
For the data set I was working with it really was no complexity at all. It certainly isn't the biggest reduction of the data but it's a fairly quick reduction to do. You could certainly partition it better but this was a quick and dirty way to do it that worked for my application.