otbutz

> Firefox supports it fully if using DoH Also for system DNS since Firefox 129: https://www.mozilla.org/en-US/firefox/129.0/releasenotes/

Wouldn't that be a positive side effect judging from the comments? https://github.com/graphhopper/graphhopper/blob/ff8f2b23a478bc17eecd409a46a35e4677e75172/core/src/main/java/com/graphhopper/routing/lm/LMApproximator.java#L123-L125

We could switch to a two-pass calculation based on the result of the plane distance. But I'm not sure we want that complexity.

Benchmarked with Germany: ## DistanceCalcEarth (master) ``` "routingLM12.air_distance_mean" : 316376.94, "routingLM12.alternative_rate" : 0.0, "routingLM12.distance_max" : 914021, "routingLM12.distance_mean" : 416693.53, "routingLM12.distance_min" : 53599, "routingLM12.failed_count" : 0, "routingLM12.guessed_algorithm" : "astarbi", "routingLM12.max" :...

That's roughly a speedup of 11% if I'm interpreting the results correctly.

> No, I think that this would be too controversial as it would allow bicycles also on `highway=path`, `highway=footway` or `highway=path` in combination with `bicycle=no`. This would allow bicycles basically...

> e.g. on openstreetmap.org there are no route hints and most mappers probably use this. Could we further avoid this? A bit of a slippery slope, isn't it? Following that...

* `NO` * `IMPLICIT` -> acting as a pedestrian * `TRAFFIC_SIGN` -> `bicycle=dismount`, `vehicle=no` > * TRUE_BICYCLE_NO (covering tag `bicycle=no`, but allowed for foot) If we're not allowed to use...

> Or do you mean this is OK because we have the `bike_access` encoded value anyway in addition and this has higher importance? Exactly. If access is prohibited, there's no...

> isn't this accessibility information already available via `foot_access && !bike_access`? Good point