[Open-transport] Next steps for Open Transport

[Andrew Byrd]

On 07/16/2013 02:55 PM, Peter K wrote:
>> 2. Creating a knowledge center on transit stops
> 
> Why yet another data format/storage and not using OpenStreetMap for
> that? It is already very easy to contribute to OSM even from mobile
> devices etc
> 
> And I fear you won't get the big enough community for that undertaking.

We should mention the other purpose of such a system besides proper
transfer times and wheelchair access information. We need to merge feeds
from different transport operators using incompatible identifiers, and
also merge those feeds with street data. Most countries do not maintain
a unified database of stop / platform identifiers (though the UK does).
We can reasonably expect that in most places such a database will not be
set up in the near future, so we might as well start building it ourselves.

We would of course fall back on distance or OSM connectivity
calculations if there is no specific information about a given transfer.

It would of course take a long time to fill up such a database, but I
think you'd also have a hard time building a big enough community to
properly micro-map all of the world's public transport stations. In many
places the result would be an enormous tangle of three-dimensional ways
for which we might never know the correct paths (I'm thinking of the
Nation or Chatelet stations in Paris).

>> what is the walking distance to get the train if you arrive by bus, ...
> 
> What if it is a big station? Or if there are several close bus stations
> too? And what is 'close'? ... the distances will be very different ...
> 
> I don't think you can and should store that in the data. You have to
> calculate it.

Why would the size of the station be a problem? Do you mean that large
complexes would lead to N^2 transfer entries, so we'd be better off
mapping all the stairs, hallways, etc. and finding shortest paths?

We clearly do not /have/ to calculate it. Many journey planning systems
simply depend on an enumeration of all possible transfers in the system.
I do see the point that in some cases a graph (OSM ways?) may be
preferrable. A combination of both techniques is likely to yield the
best results.

-Andrew