[OSM-talk] Re: One way streets

[Etienne]

On 5/16/06, Christopher Schmidt <crschmidt at crschmidt.net> wrote:
> On Tue, May 16, 2006 at 09:54:32PM +0200, Raphael Jacquot wrote:
> > Richard Fairhurst wrote:
> > >Raphael wrote:
> > >
> > >>that's totally unuseable from a machine's point. it has no
> > >>understanding of north or south or whatever. all it knows is, we have
> > >>a segment with 2 points, and you can go from a to b and / or from b to a
> > >
> > >
> > >if segment.node1.long > segment.node2.long { direction=east }
> > >else     { direction=west }
> > >
> > >...surely?
> > >
> > >I'm tempted to agree with Oliver. The danger of having 'one way'
> > >dependent on the direction of the segment is that, if a segment is
> > >accidentally reversed, the data becomes wrong. Having 'oneway=NE', or
> > >whatever, is independent of this.
>
> > besides, this is totally stupid for everywhere else where streets are
> > not in a nice grid setting. like if a street describes a nice large
> > curve, it won't work
>
> So long as a curve is always heading in a northern direction, it
> certainly does.
>
> \
>  |
> /
>
> It would mean that you couldn't desecribe a one way circle this way:
>   _
>  / \
> |   |
>  \_/
>
> but certainly a one way circle is probably going to be a relatively
> uncommon case in roads...
M25?
Lots of residential streets have loops...