[OSM-dev] SoC project idea: Wikifikation of OSM

[Steve Hill]

On Mon, 31 Mar 2008, Karl Newman wrote:

> Well, that sounds like a nice idea, but it's fraught with peril. You
> wouldn't be able to arbitrarily select any given changeset/commit for
> reversion because of edits that may have happened after that commit.

There would need to be some kind of conflict resolution strategy, much 
like Subversion has.  It would be complex, but not impossible - it would 
probably need a good user interface though (something integrated into 
JOSM?) for manual resolution of conflicts.

> For example, if someone deleted a bunch of relations, then later someone 
> else deleted a way that was a member of one of the relations, you 
> couldn't simply un-delete the relations because then you'd have a 
> referential integrity problem. Or a more subtle example--someone deletes 
> a way which shares a node with another way. Then the (formerly) shared 
> node is moved. If you un-delete the first way, is it the same as before 
> the delete? From the perspective of the database way table, it is (the 
> node list is the same), but on the map it may be drastically different.

There is no automatic solution to these problems, but they can be detected 
and the user informed.  The user can then decide what action to take 
(which could be to abort the whole thing, recreate deleted/moved nodes, 
etc).

  - Steve
    xmpp:steve at nexusuk.org   sip:steve at nexusuk.org   http://www.nexusuk.org/

      Servatis a periculum, servatis a maleficum - Whisper, Evanescence