[OSM-talk] Problems with names with postfixes / types

[Richard Weait]

2010/8/3 Brian Quinion <openstreetmap at brian.quinion.co.uk>:
> Hi,
>
> I've been handed a search bug regarding being unable to search for 福岡
> (Fukuoka) [1].  This has been caused by all cities in Japan being
> entered in the form 福岡市 (Fukuoka City).  I would normally regard this
> as a tagging error because the fact that it is a city is defined by
> the place=city tag - but in this case it is country wide [2]
>
> There has also been a recent increase in this happening in other
> places.  For example places in the UK like "Leverton CP" [3] where CP
> standard for Civil Parish, and a few relations in Germany now tagged 'xxx
> Stadt'.  There are probably other examples I haven't noticed.
>
> Up till now it has been my opinion that all items entered with the the
> type in the name were in error and I have been marking all such items
> 'will not fix' and telling people to fix the data instead - but maybe
> this is just my opinion rather than the correct decision.
>
> Anyway I'd like to promote a discussion on the topic and hopefully
> reach some sort of consensus.
>
> Options?

I found that bug report interesting as I had never considered the
significance of code pages and word splitting.

As test cases, I looked at Quebec City, Quebec  (30915641) and New
York City, New York (61785451).  Both can be found with multiple
searches.  Both are in the OSM data base as their simple name= without
the "city" suffix.

All of these work:
quebec - multiple results
quebec city - exact result - perhaps because name:en = "Quebec City"
quebec state - no results (did work once.  Did this change while I was testing?)

new york - multiple results
new york city - finds admin boundary, "more" finds city as well
new york state - exact result