[OSM-talk] Brainstorming: Simple Revert-Tools

[Gary68]

hi.

1.) with great power comes great responsibility!
2.) the number of votes should depend on the complexity (number of
affected objects?)
3.) mailing lists might be flooded because every single (and simple)
revert request is sent to a complete list - of course I don't know how
many reverts might be requested per time

generally it's an idea worth thinking about it because right now only a
few people are able to conduct such complex operations properly - but
remember 1.) 

gary68



On Wed, 2009-09-02 at 20:23 +0200, Peter Körner wrote:
> > I'd like to do a Brainstorming about how a Revert-Tool could look like, 
> > that is more open to the Community, can be used without programming 
> > knowledge and is able to to reverts fast.
> 
> I'm thinking of a process like this:
> 
> - Identify the Changeset you'd like to be reverted.
> - Go to <tool> and throw in the Changeset-ID
> - <tool> downloads the Changeset and the current state of all members
> - <tool> shows you a list of all members of the Changeset
>    - highlight conflicting changes (tag- or position-mismatch)
>    - highlight conflicts that could be reverted automatically
>      (e.g. in the malicious changeset highway=secondary was changed
>       to highway=track and on the current node it's highway=secondary
>       again, or the node/way added in the malicious changeset was deleted
>       already)
>    - propose actions on nodes/ways that must be edited by hand (like
>      jsom does when connecting two ways with conflicting tags)
> - when all conflicts are resolved <tool> generates a voting-url
> - post this url to the appropriate mailing-list (global and local) and
>    let the community vote for your revert-proposal
>    - we'll need some kind of authentication here
> - when 100 (20?, 50?, 1000?) people said "yes" to your proposal, the
>    tool applies your revert
>    - if this produces further conflicts the author should be able to
>      correct them (and only them!) without another vote.
> - there should be a history when who reverted what
> - each revert should have an explanation with a minimal length
>    (e.g. 30 words)
> 
> Please plug your own thoughts in :)
> Peter
> 
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk