[Geocoding] Broken admin boundary handling / Building centroid vs. street point

[Florian Lohoff]

On Wed, May 27, 2015 at 11:54:29AM +0100, Sarah Hoffmann wrote:
> Hi,
> 
> On Wed, May 27, 2015 at 11:43:17AM +0200, Florian Lohoff wrote:
> > i was trying to geocode those 2 Adresses and failed:
> > 
> > 	53347 Alfter (Witterschlick), Witterschlicker Allee 4
> > 		http://www.openstreetmap.org/way/89681300
> > 	53347 Alfter (Witterschlick), Witterschlicker Allee 6
> > 		http://www.openstreetmap.org/way/89678475		
> > 
> > They will be found as 53347 Bonn, Witterschlicker Allee. The problem
> > here seems to be that the building is actually in Alfter whereas the
> > street belongs to Bonn.
> > 
> > It seems not the building centroids position relative to the admin
> > boundary is relevant but the contact point on the street.
> > 
> > Is that correct?
> 
> Yes, that is correct. The address is computed entirely from
> the street.
> 
> Well, actually there is an exception for the postcode, which
> is taken from the addr:postcode if there is one. Something
> similar will happen with the other addr:* tags at some point
> in the not so immediate future.

You mean also interpreting addr:street and not using the "geometrically
next street"?

Thats my second problem which crops up currently. The geometrically next
street is not the one the building belongs to.
My guess is that this affects ~0.5-1% of all addresses. Especially on
street crossings.

Flo
-- 
Florian Lohoff                                                 f at zz.de
     We need to self-defense - GnuPG/PGP enable your email today!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: Digital signature
URL: <http://lists.openstreetmap.org/pipermail/geocoding/attachments/20150527/8b74d419/attachment.sig>