I was initially impressed with the German example of area mapping but I have had a change of heart. While an interesting experiment, and relatively well implemented in the small test area, I just don't think area mapping of ordinary roads makes sense.
To do area mapping without also doing the traditional OSM vector mapping of those roads just seems like low-grade vandalism to me. Why would a mapper choose to say, "I'm going to make a really detailed representation of road width and corner radii, that looks great on one renderer at one zoom level, and I just don't care that it breaks routing, breaks street names, and takes my time away from mapping other roads, or addresses, or crosswalks." I don't get it. It seems a very limited view of the map for one specific, perhaps selfish implementation. I expect to find wide variation in what individual mappers find worthy of their mapping time. When you consider that we have a map that we can improve with roads, intersection, interchanges, rivers and other waterways, cycle and multi-use paths, snowmobile trails, kayak routes, canoe portages, trees, steps, power pylons and lines, turn restrictions, businesses, buildings, zoos with penguin enclosures, street lighting, fire hydrants, and so many other things *deep breath*. Some will map half of these things, and many will map a much smaller subset. I understand why some mappers do cycle trails and others do coffee shops and bowling alleys. How much does the width of one road, plus the radius of the curb at the junction really add to the map? And this is all while ignoring so many other features. And as far as I can tell, these appeals to show reality more accurately extend only as far as the paved driving surface. Even the curbs are ignored. No curb:height or curb:width? Not even any indication of curb ramps, crosswalks, or audible crossing assistance? The focus is just on this idea that the curb has a radius. Students at the University of Maryland have even built a pedestrian routing system that allows choosing sloped curbs, and avoids steep inclines. http://seamster.cs.umd.edu:8090/map/index.html# Check the data, they don't bother with area-mapped roads. http://www.openstreetmap.org/?lat=38.987313&lon=-76.941263&zoom=18&layers=B000FTTT Having said all of this, and this email is too long to read, corners with curb radii can look nice. Why not put energy into allowing a renderer to draw the radius for you? Surely this can be abstracted in a way that creates a sensible corner for a large class of general intersections? Why tag and draw every blade of grass, when we can create a polygon of natural=grass? _______________________________________________ talk mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk
