[Tagging] My "weirdly unnatural aversion to relations"

Martin Koppenhoefer dieterdreist at gmail.com
Tue Oct 2 09:27:01 UTC 2018

Am Di., 2. Okt. 2018 um 06:32 Uhr schrieb Joseph Eisenberg <
joseph.eisenberg at gmail.com>:

> You are supposed to be able to get all that information by using a closed
> way for the boundary of the RV park. Then all the other features are
> contained within that boundary.

yes, and from my understanding, in these cases where the site can be
represented with a polygon, it would be better not to use a site relation
(but for many actual site relations it has been done anyway).

1. What the site relation promises to offer is adding things that are
outside a perimeter and are not polygons, to the site, i.e. nodes, linear
ways and other relations.

2. Or places, where a perimeter cannot be used, because not everything
inside it is part of the feature. You cannot "subtract" POI nodes from a
multipolygon relation (but you can not add them to a site relation).

3. It also possibly offers special roles for the members (this is the
parking for the staff of this feature, this is the ticket office for this
feature, this is the toilet for visitors of this place, ...). (But there
are no established semantical conventions how to do it).

As there is no support in osm-carto/mapnik since ever, which would probably
have channeled/unified actual usage, there are now a lot of things mapped
as site relations, that work differently. Everybody has dreamt her/his own
"jack of all trades relation" into the site relation, and quite a few of
them could be represented by a polygon without loosing semantics. That's
why I can partly agree that these relations are not "usable" (you can still
try o make some sense of them, and you will probably get more information
out of OSM if you do not ignore them).

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20181002/8b7b6c1a/attachment-0001.html>

More information about the Tagging mailing list