Re: [OSM-talk] FW: [OpenStreetMap] #1327: Rendering non-existent data

2008-11-15 Thread Jon Burgess
On Thu, 2008-11-13 at 10:40 +, Dave Stubbs wrote:
> On Thu, Nov 13, 2008 at 10:02 AM, Steve Chilton <[EMAIL PROTECTED]> wrote:
> > Trac #1327 below assigned to me but I can't resolve it.
> > It is a minor nuisance of something rendering at origin:
> > http://www.openstreetmap.org/?lat=0.59&lon=-0.16&zoom=18
> > Shows as station on mapnik, car park on cycle, station on nonames, nothing 
> > on osma
> > Editors show nothing there. Data layer says it is the API.
> > Any thoughts.
> > Just ignore it?
> >
> 
> It's because the data isn't actually /there/ -- it's somewhere else.
> When things go wrong it's not unusual for the projection code etc to
> end up sticking stuff at 0,0. The API doesn't return it though.
> 
> For example, the station is node 32009797, located at -90 latitude...
> the south pole.
> There's also a BP petrol station at the north pole, a helpful
> place=continent for Antartica, a helpful place for the North Pole, and
> a number of parking nodes inserted by osm2pgsql from parking areas
> that it obviously doesn't like.
> 
> So it's partially some bad data, and partially some odd behaviour on
> the part of osm2pgsql/postgis/mapnik.
> I'll have a go at correcting some of the data.

As Dave says, they are artefacts of the osm2pgsql conversion process. 
I will add some code to discard these when I find some time.

gis=> select osm_id,amenity,name,astext(way) from planet_osm_point where
way && setSRID('BOX3D(-10 -10, 10 10)'::box3d,900913);
  osm_id   | amenity |   name|  astext
---+-+---+---
  36966060 | | Antarctica| POINT(0 -1.5707963267949)
  32009797 | | Peñagrande| POINT(0 -1.5707963267949)
  26535074 | parking | Place de l'Eglise | POINT(0 0)
  26414180 | parking | Schüren   | POINT(0 0)
  23025831 | parking |   | POINT(0 0)
  27962310 | parking |   | POINT(0 0)
  24805472 | parking |   | POINT(0 0)
  25209770 | fuel| Station Bp| POINT(0 1.5707963267949)
 269538499 | | North Pole| POINT(0 1.5707963267949)

gis=> select osm_id,highway,ref,astext(way) from planet_osm_line where
way && setSRID('BOX3D(-10 -10, 10 10)'::box3d,900913);
  osm_id  |   highway|  ref   |  astext
--+--++---
 22529607 | unclassified | TF-463 | LINESTRING(0 1.5707963268,0 1.57)
(1 row)

Jon



___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] FW: [OpenStreetMap] #1327: Rendering non-existent data

2008-11-13 Thread Dave Stubbs
On Thu, Nov 13, 2008 at 10:02 AM, Steve Chilton <[EMAIL PROTECTED]> wrote:
> Trac #1327 below assigned to me but I can't resolve it.
> It is a minor nuisance of something rendering at origin:
> http://www.openstreetmap.org/?lat=0.59&lon=-0.16&zoom=18
> Shows as station on mapnik, car park on cycle, station on nonames, nothing on 
> osma
> Editors show nothing there. Data layer says it is the API.
> Any thoughts.
> Just ignore it?
>

It's because the data isn't actually /there/ -- it's somewhere else.
When things go wrong it's not unusual for the projection code etc to
end up sticking stuff at 0,0. The API doesn't return it though.

For example, the station is node 32009797, located at -90 latitude...
the south pole.
There's also a BP petrol station at the north pole, a helpful
place=continent for Antartica, a helpful place for the North Pole, and
a number of parking nodes inserted by osm2pgsql from parking areas
that it obviously doesn't like.

So it's partially some bad data, and partially some odd behaviour on
the part of osm2pgsql/postgis/mapnik.
I'll have a go at correcting some of the data.

Dave

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk