On Sat, Sep 19, 2009 at 11:37 PM, John Smith <span dir="ltr"><<a href="mailto:deltafoxtrot256@gmail.com">deltafoxtrot256@gmail.com</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
2009/9/20 Anthony <<a href="mailto:osm@inbox.org">osm@inbox.org</a>>:<br>
<div class="im">> This can be done without resorting to mapping each lane separately. If you<br>
> have a three lane road with no lane change restrictions or physical<br>
> barriers, you map it as one way, with three lanes, with the position as the<br>
> center of the three lanes. When the road goes to two lanes, you map it as<br>
> one way, with two lanes, with the position as the center of the two lanes.<br>
<br>
</div>I wasn't suggesting to map each lane separately, however an editor<br>
could display lanes and it would be so much better to display them as<br>
parallel ways which could be edited if they needed to be.<br>
</blockquote></div><br>That's an editor issue. If the editor wants to display lanes in a single way as parallel ways, and let you edit them if need be, it can do that.<br><br>All that's needed is an unambiguous way to represent all the various scenarios. There are some issues where I think your method of using a single way is appropriate (per lane speed when traffic is free to change lanes freely is the only one I can think of at the moment though), and some issues where I think using a single way is completely inappropriate (per lane access restrictions, per lane geometries, per lane turn restrictions). I should be able to run a shortest path algorithm on a single set of nodes and edges. Add whatever features you want to that, but don't take that away.<br>