You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
According to OpenStreetMap wiki, ways with highway=proposed and proposed=* tags are for "any feature that is still in planning phase but construction hasn't yet started". These ways are currently avoided in car profile but bicycle and foot profiles happily route on these ways even though it's possible that nothing exists on the ground.
It looks like these ways were intentionally discarded for routing on all of these profiles (car, bicycle and foot) on Jul 10, 2017. Link to merged pull request: #4258
Interestingly, routing on proposed ways was effectively re-enabled for bicycle and foot profiles about a week later on Jul 18 when a large pull request #4072 was merged. Reading through the pull request, there were no comments nor any mention in the commit message about this change. Note that the behavior was retained for car profile that had tests. It seems possible that the behavior was accidentally lost for bicycle and foot profiles during a rebase for this large pull request because these profiles did not have any tests for this behavior. taginfo.json also still says that both construction and proposed ways are discarded for routing and doesn't mention that it only applies to certain profiles.
Issue
According to OpenStreetMap wiki, ways with highway=proposed and proposed=* tags are for "any feature that is still in planning phase but construction hasn't yet started". These ways are currently avoided in car profile but bicycle and foot profiles happily route on these ways even though it's possible that nothing exists on the ground.
It looks like these ways were intentionally discarded for routing on all of these profiles (car, bicycle and foot) on Jul 10, 2017. Link to merged pull request: #4258
Interestingly, routing on proposed ways was effectively re-enabled for bicycle and foot profiles about a week later on Jul 18 when a large pull request #4072 was merged. Reading through the pull request, there were no comments nor any mention in the commit message about this change. Note that the behavior was retained for car profile that had tests. It seems possible that the behavior was accidentally lost for bicycle and foot profiles during a rebase for this large pull request because these profiles did not have any tests for this behavior. taginfo.json also still says that both construction and proposed ways are discarded for routing and doesn't mention that it only applies to certain profiles.
Steps to reproduce
At first, I noticed this problem when openstreetmap.org routed me through the following proposed way: https://www.openstreetmap.org/way/386217696
Results of further investigation are written above. Cucumber tests and a possible fix are included in pull request #6615.
Specifications
The text was updated successfully, but these errors were encountered: