[Talk-us] Talk-us Digest, Vol 79, Issue 13

[stevea]

Clifford Snow writes:
The CDP boundary seems to override the addr:city. The addr:city tag is just
the mailing address tag since the node is outside of the city limits. From
reading the documentation of Nominatim, I thought that the addr tags would
be the first choice. That's why I wonder if the problem is Nominatim.

On Wed, Jun 11, 2014 at 6:45 PM, Saikrishna Arcot <saiarcot895 at gmail.com>
wrote:

>  My expectation is that Nominatim would use the CDP (or the city region
>  it's located in) if there is no addr:city, and either
>
>     1. would allow either the CDP and addr:city when addr:city is present,
>     or
>     2. use only the addr:city if it is present.
>
>  The first option would account for the case where the addr:city is
>  incorrect (for whatever reason), but someone searching for the address
>  could either use addr:city or the CDP name.

There are multiple layers of things needing to be untangled here. 
This is (largely) the interaction of three things:

1)  CDP boundary data in OSM
2)  Nominatim's usage of 1)
3)  The as-yet-unspecified boundary between these two entities in 
ambiguous cases.

I dislike stripping it down to a more-or-less linguistic equivalence 
class as I roughly do there, but this it what it appears to be to me: 
an interaction between these two things.

What to do?  Well, how about better characterize what "we" (OSM) mean 
by including (or excluding) CDP data in our database.

Yes:  to do this well, is, roughly speaking, difficult work. 
However, we are up to the task.  I am, anyway.

This entails imagining how CDP data are, might be, and will be used 
in the future.  You can do some of that, but I don't think you can do 
all of that.  After all, OSM is founded upon the ability of "using 
them (our data) in creative, productive, or unexpected ways."  My 
sleeves have been rolled up for a couple (maybe even a few) years. 
We can do this.  It involves conversation, but we can do this.

SteveA
California