[OSM-talk] Deleted nodes on the A3

D Tucny d at tucny.com
Tue Dec 9 05:36:28 GMT 2008


2008/12/9 Richard Bullock <rb357 at cantab.net>

>
> >> >
> >>
> http://www.openstreetmap.org/?lat=51.0413&lon=-0.9116&zoom=14&layers=B000FTF
> >> > [1] http://www.openstreetmap.org/api/0.5/way/28755575
> >> > [2] http://www.openstreetmap.org/api/0.5/way/28755575/history
> >>
> >> Looking at at least one of the nodes shown in the history way
> >> http://www.openstreetmap.org/api/0.5/way/28755575/history
> >>
> >> the node display is shown as false
> >>
> >> http://www.openstreetmap.org/api/0.5/node/21076696/history
> >>
> >
> > Which means that some of the nodes were deleted from the way but the way
> > still exists...
> >
> This is exactly what happened with the M62 last month.
>
> Turns out someone had been doing lots of splitting ways to add parts to a
> Euro-route relation - and there may have been server communication errors
> when edited with Potlatch. The upshot was that one carriageway of the M62
> had become a straight line and I think something like 4 nodes remained
> (previously around 40-50).
>
> The api, data browser (i.e. www.osm.org/browse/...) and Potlatch still
> showed these extra nodes in the way (even though the nodes had been
> deleted). JOSM and the data layer showed the correct info.
>
> Interestingly, when I tried to move just one of the nodes in JOSM - and
> then
> I saved it - the api, data browser and Potlatch then reverted to the
> correct
> info (i.e. excluding the deleted nodes).
>
>
So, two solutions, if you don't want the deleted nodes, use JOSM to modify
the way and the deleted nodes will be gone, fixing the data inconsistency...
if you do want the deleted nodes in the way, use Potlatch, modify the way
and the deleted nodes will become undeleted, fixing the data
inconsistency...

Cool...

d
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20081209/191e44b3/attachment.html>


More information about the talk mailing list