Route relation interruptions

Hi there.

There are many highways with official reference that are not continuous. They go from A->B then use another highway until point C and then go from C->D.

In theses cases, the route relation should include the segment that doesn’t belong to the referenced highway (B->C)? Or not?


In the UK certainly not:

  • National Cycle Routes have been substantially revised so that all former segments which were not suitable for children have been removed. Consequently, most routes are now discontinuous.
  • A- and B-roads. There are numerous examples of discontinuous stretches of numbered highways. Some due to accidental duplication of numbers; others because part of a numbered route has been reassigned to another one.
  • some National Trails (England Coast Path) are still in development or are subject to change because of landslips etc.

It very much depends on the route network. Even within a single county within a U.S. state, the rules about discontinuous routes may differ by network. These are all common scenarios:

  • B→C belongs to both the route in question and the other route and is signposted as part of both. This is called a concurrency.
  • B→C belongs to both routes, but the route in question is not signposted. Instead, there might only be signs telling you how to get to the next leg of the route, or you might be expected to guess. This is called a hidden concurrency among roadgeeks. In the U.S., we model this scenario by including the unsignposted portion in a separate route relation, tagged unsigned_ref and joined to the rest of the route by a type=route superrelation.
  • B→C is not part of the route in question. The route has a gap.
  • B→C is not part of the route in question, but the signs imply that it is. In the U.S., I think we usually map according to the signs, but this is sometimes a source of disagreement.
  • B→C is not part of the route in question. In fact, A→B and C→D carry two distinct routes that coincidentally belong to the same network and share the same route number. Sometimes an unsignposted reference number distinguishes the two routes.

Renderers and routers that support route relations need to account for all of these scenarios.

1 Like

Thank you for your answers.
@Minh_Nguyen describes this problem very well, since those scenarios already came to my mind and that’s why I decided to open this topic.
The point 2 is the most common, I guess. The B→C stretch is not referenced with the same reference as in A→B and C→D, but has signposts indicating the direction to the C→D portion of the route (with names or references).
Nevertheless, the road routes in the official documents of the institutions which govern highways in Portugal are described with places.
For example: Road XYZ goes from Yellow Town to Brown Town and then from Black Town to Green Town.
This makes me think that this route relations should go with “B→C is not part of the route in question. The route has a gap.”

1 Like

Similar question here, we’ve got Road route reference SS81 going north-south and SS80 going west east. The two use a shortish shared east-west section in Teramo town. It used to be SS81 was mapped gapless and SS80 had a gap. Then someone changed that to the SS80 being gapless in OSM and SS81 now showing a gap there But both are using that intersection and both are document as gapless so I think that section should have ref=SS80;SS81. And the funny is, the KM posts are set as if there is no gap. That was not mapping what’s on the ground to me.

(Oh and Big G signs both route refs on that intersection too, and the Nat road manager ANAS signs both.


On my ToDo list.