On Mon, Feb 25, 2019 at 9:00 PM Sergio Manzi <s...@smz.it> wrote: > > +1 here too, and a little bit of the same concerns expressed by Andy > (https://xkcd.com/927/) > > BTW, in the Italian mailing list there is currently a thread discussing if > and how we should tag highways according to what are the official categories > in the Italian Traffic Code (Codice della Strada) are. > > There the concern is most about how to tag an official classification > (something that is implicit in the tag value in UK, if I'm not mistaken) > instead of a "descriptive classification". > > But other concerns are emerging too (at least in my head!), like the > administrative responsibility under which a given road falls (state, region, > province, municipality, private) and ad-hoc values as input for the router > (speed limits, traffic density, etc. *OR* a comprehensive "preference"value > ).
I think the official categories in Codice della Strada should probably be assigned to OSM's classes by closely matching the descriptions in the wiki. This would help Italians make sense of the map. For years Brazil's Transit code categories have been assigned to OSM's classes at least in urban areas and the result has been quite satisfactory. We do have some lack of consensus on what to do outside those areas, as OSM has many more classes to choose from that our Transit Code describes, but we're still working on it. Except for the official categories in Codice della Strada, I think such attributes of the road should go into specific tags. For example, in Brazil, the administrative responsibility is represented by the road's reference code, so it can be easily identified from ref=* tags (only the municipal level has no codes, with a few exceptions). Private roads can simply be tagged with access=private; this is common in Brazil in gated communities [1]. Speed limits can be easily specified with maxspeed=*. Traffic density is related to two things in Brazil: the road's planned function, and its de facto function. Both at urban and rural levels, many roads are not as developed as their plan would imply - not too different from the unpaved roads of Canada (still classified as trunk), which appear far away from the more developed urban centres. It seems that, not only in Brazil, it is common practice to expand a road to its final physical design only when rising demand justifies the investment. When there are no speed limit signs, it is this planned (not the de facto) function that counts legally when assessing speed limits and access rights. This leads to a situation where many roads seem unimportant when context is ignored, despite being the main routes between important, nearby places: - main routes between capitals of adjacent states being partially unpaved - main routes between a larger city and its satellite villages accessible only by unpaved roads - unpaved urban collectors (rarely, even arterials) in less developed cities - roads with few lanes that are in good condition, are high speed but have little traffic "Preference" as in "this is a good road to drive on" is very subjective. When using routers such as OSRM and GraphHopper, route choice ignores highway classification after maxspeed=* and surface=* are mapped, with the exception of access rights of particular modes (bicycles on motorways, cars on pedestrian ways, etc.). When using simpler routers based on heuristics, it is important that each network level (trunk, primary, secondary) do not have any gaps (eg. a trunk/primary through road should connect both ends of a city), otherwise computed routes are likely to needlessly tell the driver to go around usually fine routes. Finally, there's the visual aspect for users that are not using automated routers (like those using printed maps). Commercial maps do not have gaps in classification, as this makes reading harder and leads to confusion. Regards, [1] Gated community example with private residential ways: https://www.openstreetmap.org/relation/7332391 -- Fernando Trebien _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
