As the original proposer of the destination_sign I think this is a great example when the relation is useful. I assume there is some roadsign before exit 91 that announces the upcoming exit. The destination= of the relation would be what it says on the sign, which could be a place or a road number.
It has been discussed weather the ref should be the ref of the destination road after the junction. I belive there was no conclusion and ref= is not specified on the wiki page. (And has only about 40 usages in destination_sign-relations worldwide) So this exit [0] could have destination=51; Rothschield; Schofeild and I guess that you could use ref=185 specifying the exit number. [0]: http://libertymaven.com/wp-content/uploads/highway_sign_misspellings_12719.png Back at the RFC stage it was said that for signs with multiple destinations multiple relations should be created, but my opinion now is that a semicolon separated list is as good. (Or even better as it keeps us from getting hideous amounts of relations in complex crossings.) In your case you would get 3 relations. I-70 to exit 91 (with destination= according to the sign at I-70), exit 91 to exit 91A, and exit 91 to exit 91B (with destination= accoring to the sign(s) at exit 91 before the 91A/B fork). Konrad 2010/4/22 Phil! Gold <phi...@pobox.com>: > I've been trying to figure out how to tag some moderately complex exit > ramp systems when a single motorway_junction doesn't seem adequate. An > email to the newbies list led to an IRC discussion that led to a solution > that I like, so I wanted to present it here to see what other people think > and possibly what other people have done in similar situations. > > As background, most highway exits are simple and a motorway_junction with > a ref= for the exit number captures all the important information about > the exit. That approach works best, though, if there's only one > destination from the exit. I know several places where the reality is > slightly more complicated than that. One example is the exit from > westbound I-70 in Maryland (in the US) onto I-695[0]; there is one ramp > from I-70 with exit number 91 that splits a short while later into exits > 91A (I-695 southbound) and 91B (I-695 northbound). I want to be able to > capture all of that information in OpenStreetMap so that a theoretical > routing agent can say, e.g., "Take exit 91 to the right, then take exit > 91A to the left." > > [0]: I-70 westbound exit 91 http://osm.org/go/ZZd8vPYT > > The IRC discussion pointed out that this falls more or less into the same > domain as the destination_sign relation. The only thing it's missing is > the exit number. What I think would work would be to use the > destination_sign relation with the destination= tag containing the target > road (I-695, in my example), possibly extended with other destinations > from the signs in order of increasing distance (so exit 91A might have > "destination=I-695; Glen Burnie; Washington"); and the ref= tag containing > the exit number (so exit 91A would have "ref=91A"). > > I'd like to know what other people think about this idea. What other > approaches have people used for my situation? One person on IRC suggested > that ref= on the destination_sign relation sould be the ref of the > destination road, but I think it makes more sense to have it as the exit > number (perhaps there could be a destination_ref= tag). Are there any > opinions one way or the other on that? > > -- > ...computer contrarian of the first order... / http://aperiodic.net/phil/ > PGP: 026A27F2 print: D200 5BDB FC4B B24A 9248 9F7A 4322 2D22 026A 27F2 > --- -- > Steal this tagline. I did. > ---- --- -- > > _______________________________________________ > talk mailing list > talk@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk > _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk