[OSM-talk] bulk merge of duplicate points

[Elizabeth Dodd]

On Thu, 29 Apr 2010, Robin Paulson wrote:
> On 29 April 2010 21:29, Liz <edodd at billiau.net> wrote:
> > It would help if you thought a little more about duplicate things on the
> > one position.
> > E.g. Service Station
> > could have 5 or 6 types of fuel, an address, a phone number, notes about
> > ancillary services - car wash, groceries, telephone, atm machine,
> > A number of these are "amenity=xxx" and if you merge the nodes you will
> > have data loss, or difficulty in post-processing data if you string it
> > all together.
> > The next category are things that are at different heights in the same
> > location - a multistorey building with basement parking, ground level
> > shops, upper levels offices and on the roof various aerials.
> > How will you merge all of those and not lose information?
> 
> yes, we've thought about that, and realise that some of them should
> stay unmerged. the specific example we have from the import so far
> (there will be more) are fences round the edge of golf courses. the
> fence and the golf course edge do coincide, but the import scripts
> we're using have created duplicate nodes, as the two imports are from
> two different shape files within the same data set
> 

In particular, where roads in Australia overlie imported admin_boundary nodes 
and ways, we do not want them merged. Editing the relations of the 
admin_boundaries often breaks the relations. Secondly, the roads move (in 
terms of the passage of years) in rural areas, while the boundaries do not.