autobahn-bilder.de hat eine Erlaubnis für OSM erteilt, siehe http://wiki.openstreetmap.org/wiki/DE:Permissions

OSMTracker for Android is also worth looking at.

You might also do something like this using Osmarender. Relying on XSLT you can render any OSM-XML given that the XML references the Osmarender-XSLT-Stylesheet. Still, have a look at Kothic for a more current, less historic approach ;-)

Hello Dietmar,

I have just tried my hands on writing a quick an dirty Python script today. It fetches the data via the Overpass-API and compares it with preprocessed data from Tirol, Austria.

Running the comparison (including fetching the data via Overpass) takes around 5 to 20 seconds per municipality. I cannot provide nice numbers with variance and average since I am on a very slow and flaky internet connection at the moment. Actual CPU-time needed for the comparison is around 0.1s. So it is actually rather a question of how much concurrent requests the overpass-api can stand than a question of the load caused by the pre-processing and comparison of the data.

The script is not feature-complete, of course. It already takes into account associatedStreet-relations if there is not addr:street on the way/node and processes simple ranges. The code relies on serialized (pickeled) python dicts for the public data and is missing quite some features. This was okay for the quick prototype for testing the performance. In order to provide the same featureset as the current version in a robust and maintainable manner, using a database is obviously the way to go.

Considering these results I consider the disk space and maintenance-overhead of processing the whole planet and/or regular diffs overkill. Same goes for implementing a distributed system as you suggested.

I also looked at the OSM Inspector source and considering the above I found some of the algorithms worth taking into consideration.

If you are interested and if time allows, I can try to put together an early prototype and publish it.

Andreas

*had lived

Great!I have lived in Bristol for only 3,5 months and enjoyed it a lot. So it is really nice to see all those buildings appearing with such a level of detail.

Keep up the good work!