[Routing] propose a new data structure for routing

[Frederik Ramm]

Hi,

> In order to do routing with OSM data, we need to have routing nodes  
> and graph data in osm database.

No.
> Right now, osm dos not have any properly way to model or capture  
> routing information. So each and every routing applications have  
> its own way to implement routing solutions.
>

OSM does attempt to capture routing-relevant information in its data  
model. What you're proposing is the first step from a general data  
model to a routing-optimized data model. The optimizations one would  
make are very dependent on what kind of routing software you want to  
write - useful for one software, useless for another. There's  
absolutely no reason to have this in the OSM data model proper. Write  
a converter from "OSM proper" to "OSM with routing info", and then  
routing software that needs it can use this.

> Why not standardlize routing information in one place and every  
> routing applications refer to the same place? Right now, OSM can  
> not calculate routing by itself.

OSM is for data collection, not for routing; routing software is for  
routing. We will not attempt to tell the writers of routing software  
what kind of optimizations they should do; your suggestions are only  
ONE of a big number of possible ways to augment OSM data with routing  
information. I would much rather have routing software developers  
find the way that suits THEM best and pre-process OSM data accordingly.

> Routing information should be part of OSM as a whole. The main  
> purpose of OSM is to model the map in the real life and anything  
> belong to the map.
>
Yes, we need original information about where turn restrictions are,  
about maxspeeds and all that. What we don't need in the data base is  
duplicated information about whether a certain node is a routing node  
or some of the other things you have suggested; these are clearly  
derived data that should be computed in a second step.

> If OSM does not have proper way to capture routing information then  
> its data model is not completed.
>

You are only proposing to store DERIVED data in the data model, like  
estimated travel times. Such data is not "routing information  
captured", it is just pre-computed stuff that can be pre-computed by  
anyone outside of the OSM core database. No need to burden the core  
database with it.

So far I am not convinced at all by your argument. May existing  
routing applications today simply have a pre-processing step that  
does exactly what they need (instead of what someone else thinks they  
might need). Seems like a working concept to me.


Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'