This is an interesting proposal, but it needs a lot of thinking before we can consider it a working proposal.
Here is my lost of *additional points* that need clarification. 1) The problem exists in the same way also for other routes like: route=road|foot|hiking|bus|trolleybus|tram|mtb| So the wording has to be reviewed under this aspect 2) There is an unresolved issue with bicycle routes, i.e. that, variants apart, in most cases a cycle route between A and B uses a different chain of ways different in the direction A > B form the route B > A (due to things, like roundabouts, one-way streets, one-way cycle paths that are mapped as separate ways in OSM, and possible others). The only logically clean solution is to have two separate relations for the two directions under a super-relation. But none of the (many) cycle route relations I know follow this model. The usual approach is to include all ways in the same relation and use the forward|backward roles on the ways that differ. The consequences for the hierarchy model need to be looked at. 3) There is a tendency to include signposts in the relations. This has a number of open issues of its own. 3a) with the one-relation-per-direction scheme the signposts that have information for both directions, do they belong to the super-relation or do they have to be included in both child relations? 3b) In case of more routes sharing the same string of ways, many signs (referring to different routes) share the same physical support, i.e. they should be tagged on the same node. Volker Padova, Italy
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
