> This seems like a particularly strange edge case for the address scheme,
> but I'm curious if any of those valid addresses are consider the "primary?"  

I am curious too, and there is no answer actually. Ofc. whe can chose one of 
them as primary, but in most cases it's not obvious.
For example, residents may have one address in their official documents, and 
when you want to sand a mail you will threat
this address as primary, but POIs might use the second one.

> And if that's the case, what's wrong with creating a node on the building for 
> each additional valid address?  
> People looking for an amenity could look up closest POIs after finding a 
> secondary address.

It solves a half of a problem actually.
It solves only "give me coordinates by address" problem.
It doesn't solves "give me and address/addresses for coordinates or building or 
POI"

So, yes, you could find a poi using nodes, but you unable to create yellow 
pages booklet for instance.

You can't say, should other points be threated as stand-alone objects or they 
are here to indicate
that building has other addresses.

Point's could be an drawback to help solve direct geocoding problem, 
but problem of ambivalence of addresses still there.

Mon, 19 Jan 2015 05:06:23 -0600 от Paul Johnson <ba...@ursamundi.org>:
>On Mon, Jan 19, 2015 at 4:54 AM, Dmitry Kiselev  < dkise...@osm.me > wrote:
>>Andrew, you don't answered a question how to compare addr points to poi 
>>points,
>>and chose is there one address or two or more. 
>> 
>>Also there is no chance to map building with addresses with node 
>>(without polygonal geometry, in case you can't draw outline accurate enough).
>
>This seems like a particularly strange edge case for the address scheme, but 
>I'm curious if any of those valid addresses are consider the "primary?"  And 
>if that's the case, what's wrong with creating a node on the building for each 
>additional valid address?  People looking for an amenity could look up closest 
>POIs after finding a secondary address.  It's not a clean situation, but it 
>does have a couple advantages:
>*  Works with existing data consumers
>*  Simple for users to tag.
>Granted, it does have the obvious drawback you mentioned.
>
>>> This problem is already solved by bare address nodes. There's
>>> absolutely no need to introduce an additional ugly complex scheme just
>>> to avoid spatial lookups into the database.
>>
>>Please answer the questions before making such pathetic statements. 
>
> Guys, guys, there's got to be an civil way to discuss this.  We can all fight 
>when we're drunk at SOTM.  *ducks!*
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Reply via email to