The site relation was originally created for groups of features : power plant (wind turbine nodes spread over the land or sea), historical sites (often only some element (one tower, one building, ...) are historic and not the entire place) and parking (especially underground parking with only entrance mapped) spread over multiple locations. It fit exactly what is an university spread over a city or multiple places. The word "site" may be wrong, but that was the one chosen there and could be changed i suppose (i don't care for the word used myself ^_^). But creating a new relation type which would be with the same specification than a site relation would be a bit weird to me. It is overly complex for the usage no ? As the only interest is to have one feature in OSM that group all the university part and get all the university attribute. Is that really important that it is called "site" instead of "institution" ? :p
In any case, it would be interesting to define it correctly in the wiki so other mappers can find a "how to map" (either on site relation or a new relation if more people are in favour for that). :-) Le jeu. 6 févr. 2020 à 12:13, Martin Koppenhoefer <dieterdre...@gmail.com> a écrit : > > > sent from a phone > > > Il giorno 6 feb 2020, alle ore 11:37, Volker Schmidt <vosc...@gmail.com> > ha scritto: > > > > Sorry, Martin, but what do you do, if you have a big multi-storey > building and all you have is the door bell on the street level? Not map it? > > > that’s indeed a problem with multipolygons ;) > But you wouldn’t call something a „site“ either that is, hm, many sites, > would you? > > > > > The Tuebingen example illustrates the problem. The relation has two > nodes in a multipolygon as outer? That is not kosher either. > > > Yes, the Tübingen example is far from perfect, it was just an illustration > for a university with many locations, but there are many details that are > not kosher (e.g. the streets are not part of the campus, at least the > unrestricted, public ones, also the nurse residences could be questioned, > while the library arguably consists also of the grounds, not just the > building, etc.) > > Do we need another kind of relation? If we want an object for the > university, maybe yes. Adding just a tag like university=<official_name of > the university> on all the parts would maybe do the trick as well? It > wouldn’t allow for adding details about the university though (e.g. > start_date, alt_name, wikipedia/wikidata, website, operator, etc.). > > Either we could say, a university in or around the same town, seen on a > global level, is still a “site”, although on the local level it would be > seen as several sites. > Or we’d make a more generic kind of new relation for things that belong > together under a certain point of view (together they form an institution, > for example public city offices also belong to the same institution and are > often distributed over the city, or on a national level ministries and > agencies, there we’re doing it with admin level. For pt routes, there are > specific route master and network relations. > Time for an university relation? Or more generically a “type=institution” > relation? > > Cheers Martin > _______________________________________________ > Tagging mailing list > Tagging@openstreetmap.org > https://lists.openstreetmap.org/listinfo/tagging >
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging