[OSM-talk] Layer transitions
Lambert Carsten
lhc.osm at solcon.nl
Tue Aug 11 14:31:24 BST 2009
On Saturday 08 August 2009 14:11:21 Marc Schütz wrote:
> > > no, it's not, it's about relative order in the db.
> >
> > Fair enough. In other words, at any node which is a junction of way
> > segments with different layers (whether the segments are part of the
> > same way or different ways), the physical implication is that the
> > slope of the way changes in the close vicinity of that node.
>
> Not even that: it only says something about the relative ordering, not
> about slope.
>
> Regards, Marc
Here is just one of the many examples in Amsterdam concerning bridges:
http://maps.google.com/maps?f=q&source=s_q&hl=de&geocode=&q=bayreuth&sll=37.0625,-95.677068&sspn=59.467068,107.138672&ie=UTF8&t=k&ll=52.371356,4.897274&spn=0.001132,0.003447&z=19
All the connecting roads are more or less on the same physical 'layer'. The
roads that do not intersect with anything on another level don't have a layer
tag and are assumed layer=0. (A layer tag only has meaning relative to
something that is intersected).
Clearly the bridge starts at the (in this case 'X') junction. All the
connecting roads are (implied) layer=0.
The T-junction rule makes it necessary to add a non existing 'segment' OR add
three (for this bridge actually 6) layer=1 tags. Of course the other end of
those ways will connect to other roads. So the T-junction eastwards will also
need layer=1 tags which in turn will force the X intersection northwards to
have layer=1 tags. The result is of course that most or all of the roads will
need a layer tag. And where we finally connect Amsterdam to the rest of
Holland we have to hope that there are enough I connections so we don't endup
forcing the whole world to sit on layer=1.
The 'segment adding solution' does not only not represent the reality. Those
extra bits have hardly any room because it has to fit between the junction
and the 'natural=water' to work.
All the layer tag does is make it clear that a crossing of ways do NOT
connect/intersect and the numbering helps the renderer to make a decent
representation on a 2D drawing. It is not about slope or height and shouldn't
be.
>I think we should insert a "in most cases" into this rule.
What is the point of this?
O.K. if I delete the 'offending' :) bits in the wiki?
http://wiki.openstreetmap.org/wiki/Key:tunnel#How_to_Map
http://wiki.openstreetmap.org/wiki/Key:bridge#How_to_Map
Lambert Carsten
More information about the talk
mailing list