[OSM-talk] Topology

Andrew andrew at incanberra.com.au
Thu Apr 26 22:45:15 BST 2007


Frederik Ramm wrote:
> This is a problem we touched on in the paper as well.
>
> I agree with Artem that it would be desirable to make a distinction 
> between points that are really needed "as objects" - because they denote 
> an intersection, the beginning of a speed limit, or a phone booth - and 
> points that are just there to define a shape.
>
> Points "as objects" can enter into relationships with other objects; 
> "shape points" cannot.
>   
At least one GIS product uses the term "vertex" for points which only
define shape, and so can only connect to exactly two segments, and have
no attributes; and the term "node" for points which can connect to any
number of segments, and have attributes.  Maybe language something like
that  would make this discussion a little easier?
> Since there is no distinction between the two at the moment (i.e. what 
> you say, uninteresting-appearing points suddenly have meaning), it is 
> easy to break things (e.g. I want to change the shape of a stretch of 
> railway and accidentally move or delete a railway station). This could 
> of course be fixed at editor level (simply give a different colour to 
> nodes that have tags, assuming that having tags is what makes them 
> "interesting").
>
> So, one could say that making the distinction in the database or data 
> model is just a case of premature optimisation. I'm not sure, maybe it 
> is, but currently I like the idea of explicitly defining: This is a 
> point with a meaning (even if I don't give it meaning right now), and 
> this is "just a point".
>
> Bye
> Frederik
>
>   





More information about the talk mailing list