One way to find out the answer is to compare taginfo’s listing of projects that recognize sidewalk=* or sidewalk:both=* to those that recognize foot=*. Or you could try routing along a street lacking sidewalk=* in Web-based trip planners and mobile navigation applications. Or, if you’re conversant in Lua and other programming languages commonly used in routing software, you could inspect the source code.
The inference you’re asking about is far from universal. Neither the default OSRM profile nor the one maintained by FOSSGIS for osm.org consider sidewalk=* at all. Valhalla favors edges tagged with sidewalk=* over those that lack it, but this doesn’t automatically mean it’ll route a pedestrian over a road that has a sidewalk. After all, if the roadway is tagged access=no or access=private, the presence of a sidewalk doesn’t necessarily mean the pedestrian can go there.
Here are two random examples from my area: a sidewalk along a street on a military base and a sidewalk along a street on the sprawling campus of a certain fruit company.[1]
Examples of sidewalks closed off to everyone are harder to come by in OSM, and when it does happen, it’s usually a mistake. But I sometimes encounter this situation in the real world near construction sites, or built in anticipation of future development but not yet connecting anything.
Routers may send you down that sidewalk if your destination is behind the gates, on the assumption that you know what you’re doing. ↩︎