[Tagging] Feature Proposal - RFC - parking (redux)

Flaimo flaimo at gmail.com
Fri Mar 18 21:18:21 GMT 2011

On Fri, Mar 18, 2011 at 15:51, M∡rtin Koppenhoefer
<dieterdreist at gmail.com> wrote:
> 2011/3/18 Flaimo <flaimo at gmail.com>:
> I don't understand this. Inheritance of properties is implicit in
> multipolygon relations.

take a look at the example for the car rental:
it's about inheriting general values defined in the relations and
subrelations. i think with just a multipolygon this is not possible.
but i actually consider multipolygon relations more complicated to
understand than the site relation.

just relying on a surrounding amenity=parking area without a relation
also has another flaw: underground parking. basically nobody maps
underground parking facilities as areas with layer=-1. all of those i
have seen so far in OSM are mapped as nodes at the entrances. and that
is the problem. underground parking facilities often have more than
one entrance. right now, each entrance is interpreted as its own
parking lot. the relation would group them together to one parking

> the site relation is indeed used a lot (around 128.000 times) but IMHO
> needed only for details like where is the ticket office, where is the
> main entrance, which would be a suggested label position (render
> hint), etc. and there is no application that I know of (in contrary to
> multipoly) which takes site relations into account.

labeling position is not mentioned in the parking proposal and i also
don't think that it is important right now. it would be a nice
benefit, though. also the site relation for parking could be part of a
bigger site relation which would represent a consistent mapping stile.
that nobody interprets it is not the mappers or the mapping schemes
fault. sticking with an older and, in my opinion, less suitable
method, because map and routing programmers don't keep up with new
developments, shouldn't be the way to go. otherwise we still wouldn't
have the new public transport scheme, which also is also used a lot
and still isn't interpreted in mapnik.


More information about the Tagging mailing list