Well, in this case, the only way to know for a routing application what the cardinal direction is, is to look at the member roles. Either that our you slice the relation up even more to have separate relations for east / west / north / south, which to my mind would make for a too-convoluted relationship hierarchy. What is your thought on indicating cardinal direction in this case if not as member role?
Martijn van Exel > On Jan 22, 2017, at 5:24 AM, Paul Johnson <[email protected]> wrote: > > On Sat, Jan 21, 2017 at 6:11 PM, <[email protected] > <mailto:[email protected]>> wrote: > Regarding the use of child relations for routes, and what to do about > directional roles on beltways, I made some mapping changes to a beltway that > happens to be local to me. > > > > I took the relation for I-435[1] and “cloned” it into 2 new relations in > JOSM[2][3]. I then deleted all ways from the in the relation and added the > new relations, turning the old relation into the parent. As 435 is a beltway, > I added “(clockwise)” and “(counterclockwise)” to the new relations. Milepost > 0 on I-435 is the junction with I-35 at the southwest corner and the > mileposts increase going clockwise (and do not reset at the state line) so I > used the I-435 bridge over I-35 as my starting point. Starting there, I > organized the ways in the clockwise direction in the JOSM relation editor. > Once I had created a “loop,” I removed all the other ways from the clockwise > relation, then selected the members of the clockwise relation to remove them > from the counterclockwise relation. I then sorted out the ways for the > counterclockwise direction in the same way. > > > > I left the directional roles (i.e. “north,” “south,” “east,” and “west”) > intact to represent how the segments on 435 are signed, and changed roles > previously marked as “forward” back to directional roles. I also happened to > find that I had inadvertently left a gap in the counterclockwise direction in > the Johnson County Gateway project. I also noticed someone has previously > attempted to note the direction in the “ref” tag. I changed those as well. > > > > Aside from the fact that JOSM does not support the use of directional roles, > I think the changes should make it cleaner for future mappers. > > > > [1]: http://www.openstreetmap.org/relation/62155 > <http://www.openstreetmap.org/relation/62155> > [2]: http://www.openstreetmap.org/relation/6898835 > <http://www.openstreetmap.org/relation/6898835> > [3]: http://www.openstreetmap.org/relation/6898836 > <http://www.openstreetmap.org/relation/6898836> > > It would be easier to verify by using forward in the child relations > exclusively. Then it will validate as a loop, or it won't, and the gap > becomes immediately apparent. As tagged, most tools (JOSM included) won't > "get" it. > _______________________________________________ > Talk-us mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/talk-us
_______________________________________________ Talk-us mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-us
