yes, you are correct - if you don’t enforce it by e.g. avoiding ferries, the corresponding algorithm unfortunately doesn’t take turn restrictions into account.
Apart from that, th route via Dahmestraße is just the slowest. That is something the classic maps client doesn’t show, it just shows “00:02” for both routes.
If you have a look at the same route in the new client, then you’ll see that the “direct” route via Dahmestraße takes 2min 41s since Dahmestraße is a 30 km/h zone.
The first route disregarding turn restrictions takes 2min 09s, the second route via Binswangersteig takes 2min 26s.
Bike routing gives you the expected results, since bike max speed is lower than 30 km/h anyways.
If you enable some avoidables (even avoid-polygons) or maxspeed, it’ll respect turn restrictions, but only for fastest and recommended preference. They are disabled for shortest preference.
Thanks a lot for the explanation. That makes sense.
But just one last question: the routing algorithm does take into account when there are more turns on a route? Apart from the calculated numbers there is also some slow down needed when approaching turns so it is possible that the real world route™ (via Dahemstraße only) is the fastest.
But it is more an academic discussion