Hi Christoph, I just checked out the patch and found something that I think we should fix prior to committing.
The UsedNodeFilter uses a HashSet<Long> to track the node ids. Could you use a BitSetIdTracker or even better use the IdTracker interface and make the selection of BitSetIdTracker and ListIdTracker configurable with BitSetIdTracker as the default in a similar way to the bounding box task. A HashSet is unlikely to scale very well to large data sets (ie. a planet) and will consume enormous amounts of memory very quickly. Brett Christoph Sommer wrote: > Dear all, > > I recently needed to filter OSM data to only include motorways and trunk > roads. Finding no easy way to achieve that in a batch run, I extended > Osmosis to be able to filter ways and nodes to achieve this. Maybe the > patch is of use to someone: > <http://www.deltadevelopment.de/users/christoph/download/sommer-wayfilter01.patch> > > > Osmosis can then be run like this: > java -jar osmosis.jar --read-xml file="input.osm" --way-key-value > keyValueList="highway.motorway,highway.motorway_link,highway.trunk,highway.trunk_link" > --used-node --write-xml file="output.osm" > > It will filter ways to only output those with given key.value combinations, > then discard unused nodes. > > Regards, > > Christoph > > > _______________________________________________ > dev mailing list > [email protected] > http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev > _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
