On Thu, Aug 27, 2009 at 10:05 PM, "Marc Schütz"<schue...@gmx.net> wrote: > > Note that by requiring a junction, you make it impossible to model stop signs > don't involve a junction.
Yeah, I was thinking about this too.... You could argue that a stop sign/"requirement to stop" should be modeled not by "a way and a junction", nor by a "node on a way and the nearest junction" but by a *node on a way and a direction*. After all, stop signs are generally not double-sided. How about this...just tag a node (must be on a way) where the stop sign/line is in reality, with the following: stop:forward=yes, or stop:backward=yes, or stop=yes (for a node on a oneway=yes way, else implies the stop sign applies in both directions) Syntax negotiable, but you get the idea - the above was chosen to resemble some examples on http://wiki.openstreetmap.org/wiki/Key:access. You might prefer highway:forward=stop, but I don't. Maybe this is a good compromise - it avoids the need for a relation, but also clearly and completely describes the effect of the stop sign. _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk