[OSM-talk] Revert requests in general
ian.dees at gmail.com
Wed Aug 4 01:43:39 BST 2010
On Tue, Aug 3, 2010 at 7:37 PM, Ben Last <ben.last at nearmap.com> wrote:
> I think I remember something like that; however, the edits that we're
> submitting all come from one user (that represents NearMap) since we
> don't (and can't) require users of our site to all be registered with
> OSM. So we obviously don't want that single username to get banned.
> What I'm after is some process by which we can flag/raise changeset
> numbers ourselves and get them requested for reversion asap. If we
> could revert our own changesets that'd be good, but I don't have
> enough detail about the api 0.6 revert facilities to be sure that
> they'll do what we need, especially if the changeset is a couple of
> weeks old at the time that an issue is spotted.
Anyone can run the revert script if they want to. I think Frederik wrote it
in Perl and it's sitting somewhere in SVN. I think the community has decided
(so far, anyway) that we'd rather secure that functionality under a mild
layer of obscurity for now. Perhaps we should start that topic again: should
we have an automatic revert on OSM? It's technically feasible (but
The deeper issue, though, is not the revert tool, but the lack of history
viewing tools that would help make the decision about what to revert. How do
we know what changesets are bad? We don't currently have a good way to
access the complete history of a way, for example (including where it's
nodes were in previous revisions so that we can draw what it looked like in
the past). Perhaps that tool needs to be thought about before we think about
how to do a revert.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the talk