[OSM-dev] API 0.7+: Split node concept?

Alan Millar amillar503 at gmail.com
Tue Oct 12 21:33:02 BST 2010


> I am wondering (I wonder a lot lately ;-)) if some have already given a
> thought to the fact that nodes actually represent 2 different concepts in
> the current api:
> - a node in the geometrical sense, i.e. used to define a linestring/way
> - a POI

It is this way on purpose, because that is over-simplified.

There are plenty of uses for node-specific information within a way,
such as topology (whether they do or do not connect with another way),
whether they contain way-specific information (traffic lights,
gates/barriers, etc.), or have other additional useful attributes
besides lat/long (elevation, etc).   Therefore ways will continue to
need to include fully-functional nodes in addition to stripped-down
points.

It will complicate (not simplify) the editors and database if a
position in a way can flip-flop back and forth between a point and a
node.  It won't help users either.  And if that is all hidden from
users, then there was no advantage in introducing points.

Frankly, one of the main problems with the classic GIS
"shapefile"-style data paradigm is that it does not give you good
topological connectivity information, and therefore is inadequate for
OSM's multi-use data model.   If you think topology can be inferred or
derived from position, check out the various list archives for
discussions on duplicate node removal, powerlines, and layers.  You'll
find a lot of good information on it.

- Alan



More information about the dev mailing list