On Sunday, April 12, 2015 05:33:02 PM Greg Troxel wrote: > You may actually be right about the likelihood of correctness, and this > may lead to an expected value of < 0.1 errors per year. However, > imports changing data entered by hand is something that crosses a > cultural bright line, and I find it concerning that you're heading down > that path. I say that as someone who is usually much more on the > pro-import side. > > To stay within OSM norms, the thing to do is leave the existing data > alone, and publish a list someplace of mismatches. It's fine to write > to the person who added it and explain that there's a mismatch and ask > if they are sure.
Ok, I've made a bunch of changes to the code so that I make fewer changes to OSM. Please follow the links in the original email to see the (now updated) OSM changefile. > > The other notion in imports is to test out the process before you do > it. Have you run the conflation code against the osm database, and how > many cases are there where osm already has a charger station but the > tags dont' match? There are 127 such differences, the vast majority of which are the "name" tag. I have manually checked this differences: - The "name" tags I produce are better than the original - If they're not better, I manually adjust them - In many cases, the sucket:tesla_supercharger is different, sometimes "capacity" is different too, in all cases, my numbers match Tesla.com's - A small number of opening_hours tags are wrong in OSM - The resulting .osm file I produce has the old tags in an XML comment for convenience. - Going forward I can and will look at new conflicts. Charles _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-us
