|
Forgive me, if this question is already answered, but on a quick search, I could not find an answer that fits my issue 100%: It is possible to use the place=town/village/suburb etc. on an area or a node. In most cases in my area, both an area and a node are used at the same time. And when tagging, I also do it like this to keep the database uniform. Although it has the result, that a named place appears now twice in the search results, which is not good in my opinion. But it would make sense to use both as in my opinion:
What is the optimal way of having both benefits without disadvantages? Or is it just like eating the cake and keeping it at the same time? |
|
Follow the tagging good practices and in your case, the "One feature, one OSM element". Therefore, put the tag place either on a node or on the area, but not on both. For a village or town, the "area" might be:
thanks for the hint with the place_name. That should work out. Unfortunately, in Georgia we do not have any admin levels for villages and small suburbs / neighbourhoods. They either don't exist, or a source is not accessible. So for now, this workaround must do it. Also, in Tbilisi, historic neighbourhoods are sometimes split up between two raion boundaries, which is also a little annoying to tag. But with a polygon and place_name, I think it is fine.
(01 Mar '12, 06:49)
moszkva ter
|
|
In my point of view, the place=* tag should be used on a node. The spatial relation between a feature and the village/town can be deduced with the administrative boundary. See also this similar question (with my answer). The area is used for giving the administrative boundary, while the node give a more accurate information on the "place" center's position. In some case, there isn't a suitable boundary, but it is mainly for the hamlets, that are commonly attached to a village/town with a administrative boundary (at least in France). In this case, I don't think it is so annoying to not have a geometric inclusion of a feature of this hamlet in an area. I will now follow your advice partly. I will use the place= on the node and place_name= on the areas. The problem here in Georgia is, that towns or suburbs don't have an official boundary - or at least there is no access to a source for an official boundary. So the boundaries are added by common sense. Still it would be good to have at least rough outlines for the bigger areas.
(01 Mar '12, 06:44)
moszkva ter
|
|
Seems like there are differing views, because I tend to think that the area provides a superset of what the node provides, so the node can be deleted once the area is created (the node will often have been added by an earlyer mapper, as an approximation. Some arguments why a node isn't worth it when an area is available:
So again I'd say do not keep the node after you draw the area. If you really want to map the city center explicitly (say because the town has a very excentric shape), do so using a place=locality node or a landuse=* area and use "name=city center" instead of "name=Foobartown". Your advice somehow violates the "on-ground-rule". If a neighbourhood is not called "City Centre" officially, why add it as a name then?
(01 Mar '12, 06:46)
moszkva ter
1
Do not invent names like 'city center' if they are not used locally. Also, for marking the "center" of a place, there's ""admin_centre", as explained in Pieren's answer.
(14 May '14, 09:07)
sleske
|
|
The older question when-does-it-make-sense-to-use-place-on-an-area (with currently three answers) is quite about the same topic. |
I would like to expand this question: What about multipolygon consisting of all the residential/commercial/farming/... boundaries?
That should give precise borders of the city without creating duplicate geometries.