[OSM-talk] Re: [OSM-dev] Consensus on areas?

Erik Johansson emj at kth.se
Fri Sep 1 11:44:40 BST 2006


On 9/1/06, Etienne <80n80n at gmail.com> wrote:
> On 9/1/06, Lars Aronsson <lars at aronsson.se> wrote:
>
> > On August 28, Etienne wrote:
> >
> > > A way *is* an ordered list of nodes (sort of).  A way is defined
> > > as an ordered list of segments.
> >
> > That's not how ways are currently used on OSM.  I frequently draw
> > two parallel sequences of line segments for a motorway, and then
> > add all segments to one way, like this:
> >
> >   o---A---o---B---o---C---o---D---o
> >
> >   o---E---o---F---o---G---o---H---o
> >
> >   way: name=E4 ; class = motorway ;
> >        segments = ( A, E, B, F, C, G, D, H )
> >
> > Then when I discover there is an off-ramp connecting at the middle
> > of segment G, I remove that line segment, create a new node and
> > two new line segments, then I add these segments to the way:
> >
> >
> >   o---E---o---F---o-K-o-L--o---H---o
> >                        \
> >                         M--o
> >
> >        segments = ( A, E, B, F, C, D, H , K , L )
> >
> > Why?  Because what I want is a means to name the road and classify
> > it as a motorway.  I don't need the sequence, only the grouping.
>
>
> For naming and classification purposes the sequence is not important.  But
> it is very important for route planning software, which needs to know where
> a road starts and where it ends.  The sequence is also important for
> software that wants to render street names along the path of the way.

I've got lots of roads (three actually ;-) that looks like this:

A<<<<B>>>>C<<<<D<<<<E

the segments (A<-B) (B->C) (C<-D) (D<-E) forms a way with a specific
name, the segements all have directions therefor I put one way tag on
all of them. (A through E are all four way junctions..)

I don't believe the sequence of a way has much importance for creating
usable data. I know you need it for Osmrender, but I don't see it
being vital for route planning.

I only use ways as grouping since I often tag segments that aren't
even connected as a way.




-- 
/Erik




More information about the talk mailing list