Concerning OSM-Wiki a chain is used to prevent motorized vehicles. Why is there no default implication “motor_vehicle=no” documented?
Made a quick routing test: OSRM is blocking, Valhalla and Graphhopper are not blocking motor_vehicles on barrier=chain nodes. OSRM is also blocking pedestrians although not beeing motorized vehicles.
It’s the routers’ responsibility to implement these. Look at their issue tracker and add a request if it’s not already there.
… more specifically, the issue tracker of whoever has implemented the router. Most routers are highly customisable, and it may just be a wrong setting in a config file (or a failure to preprocess data before the router sees it).
Bollard, for instance, has a “Implies” section with default access restrictions.
Why not give guidance to whoever implements routers or their configs which default restrictions should be assumed for a chain?
I’d have assumed that a chain also blocks bicycles. I’d appreciate a note in the wiki that an explicit bicycle=no is required if there is no possibility to bypass the chains with a bike as one can see on the example picture. (In case my interpretation is correct)
the expectation is that a barrier blocks someone, and the recommendation is to add explicit tags who may or not pass a barrier.
There is no such recommendation in the wiki barrier page.
It actually says “Implied access differs by value (barrier type)”.
Even barrier=entrance has a documented implied access=yes.
It doesn’t make sense to me that a chain only blocks motorized vehicles and this statement wasn’t part of the proposal either. The proposal instead used the following text: “use this when the passage is closed by a chain. This kind of barrier can usually be crossed by pedestrians, but it might be forbidden to do so”
IMO, the most plausible default assumption for barrier=chain in the absence of explicit access tags would be access=no.
I agree with this.
sure, that’s one of the clearer questions, isn’t it? Typical expected situation is you can pass, because it is an entrance (and not a gate or similar barrier).
there is, for example “gate” and “lift_gate” say: “Combine with access=* where appropriate.”
Exactly, it’s on a case by case basis “Implied access differs by value (barrier type)” and not a global recommendation as you seem to suggest.