[OSM-talk] postcode idea

[Simon Hewison]

Andy Robinson wrote:
> This is a good point. How does Royal Mail differentiate in their postcode
> database between a block area which has a common postcode and a specific
> address which has a unique postcode. As far as I can tell everything is
> based around areas but that the reference to that area is a point, does
> anyone know any different?

In the postcode data there's a delivery point count per postcode. That's 
a count of distinct letter boxes that that postcode covers. For 
instance, my home postcode covers one entire side of my street (delivery 
point count of about 30), there's another postcode that covers the other 
side. At work, there's a postcode  that covers just one business. That 
has a delivery point count of 1.

The addresses with a delivery point count of 1 have the grid northings 
and eastings specific to that address, the ones for a street have the 
grid reference either averaged, or put at one end of the address range.

> For OSM I personally think that postal codes can fit any of the node,
> segment, way or area features in the database. We are not necessarily trying
> to replicate a postal code database but rather provide postal code
> information for OSM database features.

Agreed. There's not a chance that we'd ever collect and maintain the 
full set of postcodes anyway. The best that can be achieved is a 
best-effort.

-- 
Simon Hewison