[OSM-talk] Where came the concept of ways, segments, nodes?

Martijn van Oosterhout kleptog at gmail.com
Wed Mar 21 12:42:27 GMT 2007

On 3/21/07, Richard Fairhurst <richard at systemed.net> wrote:
> Martijn van Oosterhout wrote:
> > I don't know exactly (I wasn't there) but I think it came from looking
> > at existing datasets which mostly work on the basis of polylines. A
> > line consisting of multiple points. Now, that's terribly inefficient,
> > so you seperate out the duplication and you get what we have now.
> How's it inefficient? I'm curious. I'd have thought that, for the
> purposes of (say) retrieving a way, a polyline model would be much
> easier on the db than the current nodes/segments/ways model.

It's hard on editors. If you only have a set of polylines then if the
user moves a point, do you move all the other points on polylines that
have the same position? As any computer programmer knows, using
equality on floating point numbers is fraught with problems and should
be avoided at all costs.

It may be easier on the DB, but from the user's point of view it's terrible.

Another example, you have a road with a speed bump, right now you can
check if the speed bump is really part of that street, because that
node is one of the nodes of the segment. Remove the association and
you're left with user guessing if that  node is really part of it, or
just coincidently there. Roundabouts are another such issue.

I find interesting the posted mail from Andy that explains that the
system started without ways and they were only added later.

Have a nice day,
Martijn van Oosterhout <kleptog at gmail.com> http://svana.org/kleptog/

More information about the talk mailing list