On 30 January 2012 15:21, Richard Fairhurst <rich...@systemed.net> wrote: > andrzej zaborowski wrote: >> (I thought it is i->i+j, at least in JOSM it was up to some point) > > It is. But it's very difficult to extract that with certainty from a > non-trivial changeset. Add enough splits, and you may find i->i+j+k+l. Then > add some merges and some deletes, and you possibly have [p+i]+j and [l+p] > and an odd isolated section of k. > > Probably the only case in which you can actually check whether the user was > splitting, or creating afresh but using some of the same (agreeing) nodes, > is if they were using Potlatch 1's live mode. And I don't think that's been > good practice for a while. ;) > >> In any case if a way is an arrangement of node references + some >> tags, then if inside some changeset an arrangement of nodes and/ >> or tags is reused, as in your example, then, even if the editor's >> "split" operation wasn't used to arrive at it, for practical purposes >> the effects is the same. > > Practical purposes, sure, but not IP purposes. If we're saying that there is > IP in the sweat-of-the-brow required to create those tags or that > arrangement of nodes, then we need to know whose brow was sweaty.
As I understand, you're assuming that whether ways j+k were created from way i using a split operation is significant for IP purposes. I think it isn't, same as for practical purposes. Deleting a way and recreating it with the same tags and nodes inside a changeset shouldn't be treated as creating a new way, imho. The only thing that changed is the way Id. So I'd say you can get pretty good reliability of detecting splits, merges and copying tags from nodes to ways, my guess is below 1% error frequency. Whereas always assuming the object Id to represent the history of the object yields 100% error in those cases. Cheers _______________________________________________ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
