[OSM-dev] Release candidate for OSM binary format is in osmosis trunk.
scrosby at cs.rice.edu
Wed Sep 22 13:25:59 BST 2010
On Wed, Sep 22, 2010 at 2:17 AM, Frederik Ramm <frederik at remote.org> wrote:
> Kai Krueger wrote:
>> Easy to do - just download the .osm.pbf and run
>>> osmosis --read-bin country.osm.pbf --write-xml country.osm
> May I suggest it be run on planet.openstreetmap.org then as part of the
>> services alongside the regular full planet dump?
> Making the excerpts basically occupies a whole big machine for half a day.
> It might be possible to do it on a weekly basis on
> planet.openstreetmap.org (not my call though) but very unlikely to be run
Or a 4gb laptop in about 3-4 hours. I've been putting some work into the
mkgmap splitter. It can now split the planet into 1,200 pieces in two
I did a somewhat more brutal test on an 8gb i7 iMac and it took 4 hours to
split the planet into 20,000 pieces, but required 6700 concurrent open
It doesn't do all of what you might want right now, but it could be adapted.
First, the format the program uses for representing areas to extract is
bounding boxes, but this could be easily fixed; define a new file format,
import some is-point-in-polygon code, and done.
Second, the program doesn't guarantee it will output all member nodes of a
way/relation that cross the border of an area. (In osmosis terms, it returns
--boundingbox completeWays=false completeRelations=false.). However, it
ameliorates this by automatically expanding each bbox in question by adding
an overlap border.
Third, the input format is *.osm.gz or *.osm.pbf. The output is *.osm.gz and
without metadata (version/timestamp/user/...).
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the dev