On Wed, May 23, 2012 at 8:39 AM, Martin Vonwald <imagic....@gmail.com>wrote:
> You have to keep in mind that most of the streets are in fact a > collection of parallel features. Only at some points (junctions, ends) > this might not be true. The proposed relation might(!) be a solution > for some special cases (e.g. irregular steps), but for the rest this > is unmanageable imo. Any solution should concentrate on the common and > make it simple, but also allow to handle the exceptions - maybe with > some extra effort. This proposed solution seems to concentrate on the > exceptions. > > > +1. Streets are predominantly 1-d objects. Computers are a lot better at unpacking stuff than repacking. Yes you can write intelligent algorithms to do it (at least, I begin to see how you might do it...), but the dataset's big enough as it is, without removing a bunch of useful constraints. Now if we had an editor that displayed some parallel lines if you put in cycleway=track, and maybe something similar on the standard rendering... A fancy renderer might take further tags about the degree and nature of separation into account, perhaps even interpolating between values on nodes. That's all entirely extensible. A botched lets-just-put-it-all-in-a-relation-and-hope-someone-writes-an-algorithm-to-decipher-it is probably just creating data, and destroying information.
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
