This might be tangential to the discussion, on the other hand, it might be a
kind of "hidden" or unstated assumption about how ways are "interpreted" in OSM
to mean some implied given semantic, which in my opinion, they shouldn't do.
So it could be revealing. Here goes: any given way should be explicit as to
what it is doing there by virtue of two things, its tags and its relation
memberships (if any). A third might be "connectivity," that is, whether a
terminal node of the way is shared by another way. But, "that's it."
I think everybody can agree that there are many, many methods of OSM's data
existing in situ (in their proper place) such that all of the variations 100%
accurately describe the data. For example, whether a segment of highway
between points A and B is made up of one single way (ideal in some sense of
"perfection") or two, or three segments of way that have been split, doesn't
really matter to the underlying correctness of things. Of course, we don't
want to split that way into a needlessly large number of segments "just for
grins." But then, there ARE cases where splitting a way is exactly the right
thing to do, for example so that correct segment(s) are properly accommodated
as elements of a (route...) relation.
I don't think OSM data consumers can rely upon a given way representing whether
a legal distinction exists simply by the way being "this way" vs. "that way"
(because one way got split at "some particular location"). Let the tags on
those ways or their inclusion in a relation do that, because that is reliable
and clearly stated (by explicit tags, or explicit inclusion as members).
_______________________________________________
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au