[OSM-talk] line length; checks in server

[Joerg Ostertag (OSM Munich/Germany)]

On Thursday 17 August 2006 11:37, Nicola Ranaldo wrote:
> > > * Forbid to {add/change} a segment witch have the same start and
> > > enpoint than a already existing.
>
> forbid segement from a to b, if exists a segment from b to a ?

But only if it has the same tags. Think of:
 - Two streets above each other. One on each level of a multilevel Bridge.  
   I've seen quite some of those
 - A footpath under a bridge of a highway
 - A shopping center/car park/... with multiple lanes on top 
   each other in differnent floors.

> > This will have to do a segment lookup. We'll have to test what the
> > performance impact is. But I definitely would vote for doing this one two
> > if the perfomange impact is not too big.
>
> are there statistics about read/write api call available? i think the
> overall is map and object requests...
>
> > > * Forbid a node with the same lon and lat.
> >
> > Only if they have the same tags.
> >   - Same lat/lon, different elevation for example
>
> Yes... this has sense
>
> >   - same position two differnt PointOfInterest with different name= tags
>
> Actual scheme will suggest this, but i think you have to choose and
> formalize a robust solution becouse two nodes with the same
> lat/lon/elevations let users create unconnected paths. If we adopt the
> duplicated schema all editors should warn users of these nodes while
> selecting them.

Warn - maybe, but still allow them. I'm just thinking of Boat lifting stations 
with different elevation, elevators with exits in different levels, ...

Maybe if there are more than one point at the current location the editors 
have to ask which of those you meant to connect. Other mapping programs like  
MapSource(Garmin) also ask which attribute you meant if you click on a spot 
where more than one poit/attribute would aply.

And more examples would be POIs:
 - a Hotel, which has also a restaurant
 - a Food shop which has a Postal office inside

-

Joerg