Multiple amenities in one node

Kozuch · 2010-10-09T15:28:47.000Z

How do I solve a situation where there are more services in one place. For example Theatre and Restaurant…

MikeN1 · 2010-10-10T10:58:06.000Z

I have created 2 nodes for this situation. Trying to combine both into 1 node may not work in all cases because they use the same ‘amenity=’ tag, and some software may not properly handle 2 POIs in 1 node. Similarly, separating the POIs with semicolons in one tag will probably not work with search algorithms.

stephan75 · 2010-10-10T15:23:12.000Z

This sounds like a very good questiion for the FAQ system at help.openstreetmap.org …

maybe you could create an antry for that? (If there isn’t already one for that problem?)

Tordanik · 2010-10-10T20:50:17.000Z

How do I solve a situation where there are more services in one place. For example Theatre and Restaurant…

It somewhat depends on how the situation is in reality. Are the two features really the same entity, are they only close to each other, or is there an “is in” relationship?

If one feature is within the other one, then it’s best to map the outer one as an area, and add the inner one as a node or smaller area within that area. Software can (but, at least currently, usually doesn’t) understand the “is in” relationship expressed by this; and even if it doesn’t, it will be able to recognize the two features.
If the feature really is a single entity (which would also mean that additional tags, such as opening_hours or operator, are the same), then mapping it on a single object - node or area - makes sense. For example, a shop that belongs to different shop categories could get a shop= tag with semicolon separated values. Don’t expect this to work well with current applications, though.
If these are two independent features that happen to be in the “same” place (e.g. in the same house), but are unrelated except for proximity, then don’t even think about using the same node for them. In that case, they should clearly get separate nodes/areas.

MikeN:

I have created 2 nodes for this situation.

Software doesn’t understand this correctly either (it will rightly think that these are two entirely independent features, such as case 3 above), but for most use cases, it will indeed fail in a more desirable way. For example, duplicates within a search result list are still better than missing search results…