On Fri, Jul 08, 2011 at 08:05:47PM +0200, colliar wrote: > Am 08.07.2011 05:49, schrieb John Harvey: > > I find there are a lot more abbreviations if you look at addr:street= > > rather than the name= . I suspect that with mobile entry of POI's we > > are going to see more and more abbreviations being entered, just because > > mobile keyboards are slow. I would applaud a bot that asked me if I > > meant the nearby "Main Street" when I entered "Main St.". I would also > > applaud a bot that converted loose addresses like this into better > > structured relations like: > > > > http://wiki.openstreetmap.org/wiki/Proposed_features/House_numbers/Karlsruhe_Schema#Using_relations_to_associate_house_and_street_.28optional.29 > > > > I use associatedStreet relations for some time now but we might need to > adjust it a bit: > > 1. More than one way with role=street should be allowed. > Otherwise you end up with lots of relations and I do not know any editor > which supports this relation when splitting ways.
I use type=street: http://wiki.openstreetmap.org/wiki/Relations/Proposed/Street which in my opinion is more concise and flexible, and permits multiple ways with role=street. There are 10k instances instead of 40k for associatedStreet. Unfortunately the tools don't seem to understand it yet, e.g. searching for "19 third avenue, york", nominatum gets the wrong house: http://www.openstreetmap.org/?relation=1021672 Cheers James > 2. the role=house should also work with closed ways and relations. > For closed ways it is obviours since buildings are mapped as areas. > I found many places where an area with several buildings has one > address, sometimes theses areas are site or multipolygon relations. > > I found some streets with more than one postcode. For these streets I > used one relation for each postcode and added the postcode in the > relation's name + addr:street=[Streetname] > > I grouped them in a main relation which might be not needed. > > >> i was under the impression consensus was to type the full word, then > >> renderers would shorten where necessary? apparently some mappers > >> disagree though > > +1 > > > cu colliar > > _______________________________________________ > talk mailing list > talk@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
