vehicle parameter

The vehicle parameter was replaced by explicit statements in the custom model. So instead of:

profiles:
 - name: car
   vehicle: car
   custom_model: {
     "distance_influence": 10
   }

You write:

profiles:
 - name: car
   custom_model: {
     "priority": [
       { "if": "!car_access", "multiply_by": "0" }
     ],
     "speed": [
       { "if": "true", "limit_to": "car_average_speed" }
     ],
     "distance_influence": 10
   }

Or for vehicles with a priority encoded value you write:

profiles:
 - name: foot
   custom_model: {
     "priority": [
       { "if": "foot_access", "multiply_by": "foot_priority" },
       { "else": "", "multiply_by": "0" }
     ],
     "speed": [
       { "if": "true", "limit_to": "foot_average_speed" }
     ]
   }

This looks more verbose but for the standard vehicles we have created default custom model files. So instead of the above custom model for car you can also just write:

profiles:
 - name: car
   custom_model_files: [car.json]

With turn costs this looks now:

profiles:
 - name: car
   turn_costs:
      restrictions: [motorcar, motor_vehicle]
      u_turn_costs: 60
   custom_model_files: [car.json]

Furthermore the new approach is more flexible. A profile that previously required the special vehicle roads (to get unlimited priority and speed) always created the encoded values roads_access and roads_average_speed and used unnecessary memory and introduced unnecessary limitations. Now these artificial encoded values are no longer necessary and you could write custom models even without any encoded values and only one unconditional or if-else block is necessary for speed:

profiles:
 - name: foot
   custom_model: {
     "priority": [
       { "if": "road_class == MOTORWAY", "multiply_by": "0" }
     ],
     "speed": [
       { "if": "true", "limit_to": "6" }
     ]
   }

graph.encoded_values

All encoded values that are used in a custom models must be listed here.

If you used a property like block_private=false for e.g. the car vehicle, you can now use this property for the encoded value car_access:

  graph.encoded_values: car_access|block_private=false

shortest and fastest weighting

Both weightings were replaced by the custom model. Instead of weighting: fastest you now use the default custom weighting as explained in the previous section.

Instead of weighting: shortest you now use a custom weighting with a high distance_influence:

profiles:
 - name: car
   custom_model: {
     "priority": [
       { "if": "!car_access", "multiply_by": "0" }
     ],
     "speed": [
       { "if": "true", "limit_to": "car_average_speed" }
     ],
     "distance_influence": 200
   }

temporal conditional access restrictions

car_access, bike_access and foot_access do no longer include the conditional access restrictions by default. If you want the old behaviour e.g. for car you need to add the following statement in the priority section of your custom model:

{ "if": "car_temporal_access == NO", "multiply_by": "0" }

Depending on the use case e.g. for foot it might make more sense to use the new default and show the conditional restriction value via the new path details access_conditional, vehicle_conditional etc. built from the OSM tags access:conditional, vehicle:conditional etc. See how we utilized this for GraphHopper Maps  with a separate route hint (icon below the route distance).

max_slope

Is now a signed value. To get the previous behavious use:

Math.abs(max_slope)