|
There is a weekly database dump called Planet.osm. How is it created? The answer could include details like former anonymous edits, referential integrity, etc. |
|
It's created using a custom C program called planetdump. You can find the source here: http://trac.openstreetmap.org/browser/applications/utils/planet.osm/C It runs inside a transaction so should provide a snapshot and have the same level of referential integrity as the database. There is no such thing as a "former anonymous edit" as far as I know. Any user whose edits are public will be identified in the dump and any user whose edits are not public will not be identified. |
|
The planet dump is created by running the planet dump program against the OpenStreetMap database. The run is started at 1:11am UK local time, and the exact time is available in the timestamp attribute of the <osm> element at the beginning of the file. Planet files are compressed with pbzip2, which makes them unsuitable for some decompressors, notably the Java native bzip2 implementation. Planet files may contain ways which reference nodes which have been deleted, or relations which reference members which have been deleted. In planet files generated since March 2010 this is due to errors in the database, not with the dump process, and these errors can be found and fixed. In planet files generated before March 2010 this is due to the dump process not being run in a sufficiently isolated transaction. Planet files are in the standard OSM XML format, and only contain visible (not deleted) elements. Anonymous edits are present, but lacking the uid and user attributes. |
