That's super great! Excited to hear that. On Tue, Jun 30, 2009 at 7:34 AM, Joe Richards<joefis...@yahoo.com> wrote: > > I've been working on the LINZ data import, on attribution/legal as well > as the actual import. The LINZ data I have is actually via the NZ > Open GPS project and is in Polish Map (.mp) format, which seems > to use Garmin types for its features. > > I've hacked around with a few other scripts and have created a python > script which creates .osm files, and then imported them into a local > postGIS instance I run at home. From there I've generated tiles using > Mapnik and uploaded them to a dev server for your perusal.... > > Upload is still in progress but most are there already (starting in the > north and working south). It's just south of Christchurch already and > I have gone and rendered Dunedin, Queenstown and Glenorchy ahead > of time. If you find zoom 16 is not visible then zoom out a bit until you > find the available tiles. As mentioned all of the north island is done. > > http://linz.dev.openstreetmap.org/~JoeRichards/ > > Notes: > * import was done on a basic world map (from vmap0) to provide > coastlines where they were missing > * some large rivers have a lot of detail, but some seem to be missing > altogether (e.g. Lower Hutt river) - this was also missing in the NZOGPS > dataset, > not sure why > * tiles are still being generated and uploaded now (Tue 30th June), > but all of the north island to 16 zoom levels is done as well as the > Tasman... South of Christchurch is still being uploaded (although some > tiles are there at lower zoom levels) > * I think most of the road types (primary, secondary, trunk) etc > might be completely off, including the link roads. Please send me any > specific instances or comments on this > * No attempt made to support anything like turn restrictions or > relations since this was missing from the original dataset > * Roundabouts seem to have an issue as well, where their parts are > missing > * large lakes such as Lake Wakatipu near Queenstown were split across > two datasets, and as such as not showing up > * central cities seem to have more detail in the existing OSM data > (except perhaps some missing street names) but rural areas have > less... This is ok, but your thoughts are welcome on how to merge > these > * parcel boundaries (cadastral) information, showing property boundaries, > shapes, and house numbers will be imported separately, ie are to-do > * Department of Conservation parks, national parks, reserves etc are > also to-do, although some are present in the current dataset. > * No data is merged with current OSM data, so this is really to spot problems > before we attempt merging. It has been suggested we might use > the java-based RoadMatcher for this. Your comments are welcome. > * Thanks to Geofabrik.de for allowing me to use their "compare" javascript > > I will also put up the .osm files generated by the script so you can > check out whether the data is there, but just incorrectly tagged. > > Please feedback to me anything you see, especially if it is incorrect. > Since the plan is to merge this with the existing Openstreetmap data, > missing features that are in OSM are less of an issue than features > that should be in the imported dataset but are not (and hence are > missing on both). > > Source (GPL) is here > http://trac.openstreetmap.org/browser/applications/utils/import/linz2osm/mp2osm_linz_jr.py > > Anything in the code marked as TODO or FIXME requires special attention and > verification. > > Enjoy and feel free to comment! > > > > > > _______________________________________________ > dev mailing list > d...@openstreetmap.org > http://lists.openstreetmap.org/listinfo/dev >
-- anselm 415 215 4856 http://hook.org _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk