[OSM-talk] Inheritance of roles in nested relations

Frederik Ramm frederik at remote.org
Sat Dec 20 22:33:25 GMT 2008


Hugh Barnes wrote:
> To provide clear guidance for clients, because as you said — and I think
> is right, though I can't currently think of examples — this inheritance
> is not always desirable. Also, component=yes prevents someone
> deleting something that looks useless on its own.

If you try to delete something that is part of a relation, either your 
editor will ask you whether you want to remove it from the relation, or 
it won't and get an error back from the API.

I maintain that it is not necessary to explicitly tag the parent or 
children as the exact meaning can always be derived from the context; 
but then again a little redundancy is not bad, makes it easier to spot 

Be aware however that there is some tagging ambiguity here, for example:

Bus route relation
    -> contains many
          type "X" relations
             -> each of which contains many

Now the type "X" relation could be a grouping of ways into superways 
(someone collected all parts making up Oxford St into one relation, the 
bus goes down the length of Oxford St, so you use that relation to route 
your bus over). In that case, "X" would probably be tagged with usual 
highway attributes.

The type "X" relation could also be a sub-structure relation of the kind 
you're talking about, in which case you would expect the parent/child 
tagging scheme you were talking about.

It's all a bit murky but I presume we'll get it sorted within a few 
years ;-)


Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"

More information about the talk mailing list