Re: Zip Code Proximity - Mailing list pgsql-general

From Jeff Hoffmann
Subject Re: Zip Code Proximity
Date
Msg-id 392423F2.9DE224F9@propertykey.com
Whole thread Raw
In response to Re: Zip Code Proximity  (Andy Lewis <alewis@recruitersonline.com>)
Responses Re: Zip Code Proximity  (Andy Lewis <alewis@recruitersonline.com>)
List pgsql-general
Andy Lewis wrote:
>
> Actually I was thinking more on the lines of:
>
> select location from test where location @ '((31.6283,93.6347), 1.39)'::circle;
>

that works just great, assuming that you can accurately calculate the
radius of the circle that you're searching for.

> If Zylks, LA (lat/lon = 32.9696,93.9801) is also in the DB.
> The above query should select Zylks since 1.39 * 62.1 = 86 miles approx.
> 86 miles is about correct for those two zip codes.
>
> Wouldn't this be much easier?

to me, the big question is how you're calculating the radius & if you're
taking into account the curvature of the earth so that 62.1 changes in
different parts of the country.  if you've taken that into
consideration, great.  it's probably going to be ok to flatten the earth
for an area that size, though.

> Is the data type POINT index-able?
>

unfortunately, not in a really useful way for you.  plus, the optimizer
probably wouldn't use the index any.  it's a small data set, something
around 35k-40k records, right?

pgsql-general by date:

Previous
From: Andrew Sullivan
Date:
Subject: Re: initdb and "exit_nicely"...
Next
From: Jurgen Defurne
Date:
Subject: Re: max tuple size and filesystem block size