[OSM-talk] A post box called Breuningsweiler

David Earl david at frankieandshadow.com
Fri May 25 12:27:24 BST 2007


Andy said:

> I'm just been using out-of-copyright 1:25,000 mapping to do a test area of
> input that includes a lot of named POI data (nodes) covering the sort of
> things you see on traditional mapping, such as named farms,
> woods, coppices,
> named cottages and Halls, named collieries and all that sort of stuff. In
> most cases these seem best annotated under the "place" and "name" tags so
> I've been doing the following (examples):
>
> place=farm, name=Blah Farm
> place=building, name=Blah Hall or Blah Cottages
> place=wood, name=Blah Wood or Blah Coppice
> place=works, name=Blah Colliery (Dis)
>
> All these and other examples seem to sit better under the place tag than
> anything else, they are all there on the original map to define a place by
> its name and location and are just smaller or more specific place tags and
> in my view a logical lower extension of the city/town/village/hamlet.....

This seems perverse to me when there are already tags for farms
(landuse=farm), woods(natural=wood or landuse=forest), works
(landuse=industrial) and so on. place is intended for conurbations. Airports
shouldn't have place, but most do because whoever imported them all in bulk
did that throughout.

Both mapnik and osmarender render woods, but they need the natural=wood to
do so. Are you saying you've created areas of natural=wood and a node within
it place=wood? Even so natural=wood is defined to apply to nodes as well; in
principle either of them should be able to render the name if given, but
they won't be looking for place=wood in order to render a name (unless a
renderer sets a default name renderign style for place nodes it doesn't
recognise).

Why go against the grain and make it harder for the data consumers?

(Likewise, I've noticed quite a number of highway=minor in the Norwich area;
again why not go with the flow?)

David





More information about the talk mailing list