-
Notifications
You must be signed in to change notification settings - Fork 86
Description
The following query on the OSM data https://qlever.cs.uni-freiburg.de/osm-planet/xCerw9
SELECT ?osm_id ?geometry ?lng ?lat WHERE {
osmrel:62768 ogc:sfContains ?osm_id .
?osm_id rdf:type osm:node .
?osm_id geo:hasGeometry/geo:asWKT ?geometry .
BIND (geof:longitude(?geometry) AS ?lng) .
BIND (geof:latitude(?geometry) AS ?lat) .
}
LIMIT 3
returns the following result:
| ?osm_id | ?geometry | ?lng | ?lat |
| osmnode:10003156787 | POINT(7.854637 48.017093) | 7.85464 | 48.0171 |
| osmnode:10004601576 | POINT(7.830281 48.002005) | 7.83028 | 48.002 |
| osmnode:10005201820 | POINT(7.821752 48.003969) | 7.82175 | 48.004 |
@ullingerc Do you understand where the precision loss comes from? In my understanding the POINT(... ...)
is already rounded due to QLever's limited precision for points. But why is that rounded result rounded again when applying geof:longitude
or geof:latitude
?
PS: It seems that the precision of all our doubles is only four digits after the dot https://qlever.cs.uni-freiburg.de/wikidata/6n45iu