Richard Welty wrote:
what Martjin is after is improving the situation for routing engines that use OSM data. in the US at least, highways, regardless of their actual direction, are assigned cardinal directions and signage frequently uses those directions. the premise is that if the cardinal directions are recorded in an appropriate manner (e.g., as the roles for ways in a route relation) then they are available for the spoken directions from a routing widget of some sort.
But surely the role of a way within a relation is a different thing entirely to the cardinal direction of a relation?
For example, imagine I'm near 2 ways that are both part of the I-80[1]. For simplicity let's assume that the way that I would take to get to eastbound destinations actually goes in an eastbound direction at my location.
If it's important to recognise that there are two signed I-80 routes, one eastbound and one westbound, shouldn't there actually be two I-80 relations? The "eastboundness" is really a property of the route, rather than the individual way.
Now imagine a situation where an imaginary I-80 ran along a non-divided road for a while. There'd only be one way, and it'd be part of both relations. With normal OSM tagging, the way might have a "forward" role within the eastbound relation and a "backward" role within the westbound relation. The US-style "west" or "east" tagging fails here because the way would need to have both "west" and "east" roles.
Also, perhaps a list other than josm-dev might be a better forum for discussing how best to map route relations?
Cheers, Andy [1] Example of how a bit of the I-80 is mapped now, for info: http://www.openstreetmap.org/browse/way/10473201 http://www.openstreetmap.org/browse/way/30411369 _______________________________________________ josm-dev mailing list josm-dev@openstreetmap.org https://lists.openstreetmap.org/listinfo/josm-dev
