[Talk-transit] Tagging of railway station

[fly]

On 07.12.2013 09:47, Roland Olbricht wrote:
> Hi,
> 
>> 1. Position of tag railway=station
>> There are currently two approaches [1]:
>> (i) on a node within the main concourse area
>> (ii) on an area encompassing the land used for passenger services (including
>> any concourse, platforms and associated tracks)
> 
> I strongly opt for (i). As you have mentioned, for both the exact placment is 
> subjective. But the single node is far easier to understand and handle for 
> mappers.
> 
> For example, you could tell a mapper that the node is the location where label 
> is placed. By contrast, to start a mapper's introduction with a lengthy 
> explanation of the computation of a centroid is not practical.

This problem is not know (see place=*). We even already have an
solution: role "label".

I tend to use stop-area relations for stations.

> If you want a really precise station description, I would go towards indoor 
> mapping, see
> http://wiki.osm.org/wiki/Indoor_Mapping
> In-station navigation could be a strength of OSM if enough stations are 
> mapped.

Your comment below contradicts the one above.

>> N.B. A third approach is presented on the wiki [2]: tagging the building.
>> This approach seems not appropriate, since the bulding(s) often doesn't
>> cover the whole station surface (e.g. platform area). Maybe it could be
>> removed frome the wiki ?
> 
> Yes, remove it from the wiki. A lot of stations even don't have a building.

+1, cause the building should be tagged with building=station and the
station includes the building but goes far beyond.

>> 2. Use of public_transport=station
>> This is a much debated point:
> [...]
>> (ii) public_transport=station should be added to the one object that is
>> tagged railway=station, since these tags are synonymous
> [...]
>> Same question: is there one of these four approaches that should be favoured
>> ?
> 
> Clearly (ii) is the best choice. Even the public transport proposal states 
> that the tags should be used alongside the existing tags. In general, a mapper 
> and also a tool would usually just left aside unknown tags, so (ii) keeps 
> tools working regarless on what tagging they depend.
> 
> Having two distinct objects would be difficult to understand, because which 
> object you get will then depend on the tools syntax (either one of them or 
> both), and nobody expects a second difficult-to-find shadow object to exist.

For new created objects I only use the new scheme but I do not delete
the older tags if already tagged but only add the new ones.

Cheers fly