|
The relevant Wiki page says that the toll=* key is approved for use only on ways, and usage in Taginfo reflects this. But shouldn't adding the tag to a single node (the tollbooth) be adequate for routers to recognise that the road requires a toll? Having to explicitly mark each portion of a road where a toll is applicable seems very redundant, since this is just essentially tracing possible routes that go through the tollbooth. |
|
Like many things in OSM, there are many right answers. In one way, you're right, it could be possible to just have the toll tag on the node, and people can "figure it out" from there. In theory, you only need to pay the toll at the toll booth. However in other way, usually the law is done such that "If you want to drive on this stretch, you must pay a toll". People often refer to "toll roads" not "toll points". Regardless, the community of people who add data to the map (mappers) and people who use the data from the map are expecting it to be tagged on the way, not a node. You should do what people are looking for. |
|
We typically map the effect on the road. We do the same e.g. for maxspeed. We prefer to map that on the road instead of mapping the individual sign (although that is done as well, but more as information to other maps). In case you map it only on a node of a road where traffic goes in both directions, you have to indicate whether the toll road is left or right of the node. |
|
It's a convenience for data consumers who want to determine whether a way is a tollway directly by examining way tags rather than fetching and analyzing a potentially large network to determine that there is no exit without passing through a toll node. 2
Not just data consumers, but contributors: no-one wants to search along 10s of kilometres of road for the toll booth to check if a road is correctly mapped.
(27 Jul '15, 11:13)
SK53 ♦
Okay, makes sense. It's just quite a bit of a hassle, though, when you have to manually untangle motorway interchanges to find out which ramps may lead to an exit and which don't.
(27 Jul '15, 12:59)
Paul_012
Yeah, but much better than a mapper does this once than expecting every data consumer to do it in slightly different ways!
(28 Jul '15, 19:19)
SK53 ♦
|
