Open Data License/Backup Plan

From OpenStreetMap Wiki
Jump to: navigation, search

As part of the move to the Open Data License it will be necessary to remove or hide those edits which have not been re-licensed. Of course, it would be better not to have to remove any data, but there are likely to be some contributors who do not or cannot re-license.

Users are encouraged to engage in Remapping before the changeover date to minimise the amount of disruption caused by the process.

The proposed plan

These are the proposed stages of the migration process, subject to change based on technical or policy feasibility. This is not the final plan, and may change as we figure out the best way to do it, but this is the general outline. A date will be announced for this in advance based on the Implementation Plan.

  1. Database is taken to read-only mode.
  2. A dump of all geographic data, most likely full-history planet, is made available under CC-BY-SA. This will be hosted on [1] and mirrors and we will try to keep it available for as long as is practical. (Other sites may choose to continue to offer this data through an API or other means.)
  3. Each element is examined, and rules are applied to remove content that cannot be relicensed under the new Contributor Terms:
    • Additions and changes made by users who have not agreed to the CTs will be removed or reverted. For example, if a tag was changed in v3 by a non-agreeing user, that tag change will not be carried through to the 'retained' element; if an element was added to a route relation by a non-agreeing user, that element will not be included in the 'retained' route relation; and so on.
    • Subsequent changes derived from these additions will be removed. Note that this is not a simple task of deleting all versions after an edit by a non-agreeing user; it depends on the nature of the edit and whether subsequent changes are deemed to be derived from it.
    • The exact rules to be applied are under development.
  4. The database will be taken to offline mode.
  5. Elements and versions not marked "OK" are hidden/deleted.
  6. A "fixup" script will be run to fix errors introduced by such hiding and restore referential integrity.
  7. The database will be taken to read-only mode.
  8. Dump first ODbL planet, possibly both a full-history and current.
  9. Database is taken to read-write mode.

At no point in the process will we delete data which hasn't been made available.

Per-changeset relicensing

In general, the factor that determines whether changes are retained in the ODbL-licensed database will be whether the editing user has agreed to the Contributor Terms.

However, the option does exist for users to declare that some, but not all, of their changesets can be relicensed. This is particularly relevant for users who have imported or bulk-changed data using their own account (which is discouraged by the Import/Guidelines). Users for whom this is relevant should contact the Licensing Working Group.

This may also be relevant to the simplest mechanical ('bot') edits which are considered not to attract copyright protection.