For an extreme example, you can have a look at Europaplein 360 - 398 in Utrecht (With an editor). Our reason for this solution is that we want to make sure that the addresses stay within the building contour. The imported dataset only guarantees that the addresses are within the building. When aligning the addresses parallel to the street, the address nodes could easily end up in a neighboring building. Also you would want to make sure the address numbers get aligned in the right direction along the street. And in The Netherlands, the addresses might be in different streets. The algorithm would get quite complex if you wanted to solve all these issues.
Gertjan On Thu, 2014-10-30 at 15:47 -0500, Eric Ladner wrote: > On Wed, Oct 29, 2014 at 5:29 AM, Gertjan Idema <[email protected]> > wrote: > > Matt, > > The principle dividing up multiple addresses on the same > location within the building is quite simple. > If there are multiple addresses on one location within a > building I do the following. > > 1. Sort the addresses by postcode, street, house number. > 2. Determine the angle of the line pointing from the address > location to the center of the building. > 3. From the angle and the desired distance between the address > nodes, calculate a delta x an a delta y. Either or both may be > negative. > 4. Iterate over the address nodes and add (i * delta x) to > the x coordinate an (i * delta y) to the y coordinate. > > If the address location is at the center of the building, I > set the angle to 0. > > > > > > I'm having problems visualizing that in my head... Picture? > > > If it's doing what I think it's doing, whouldn't it make more sense to > align the address nodes parallel to the nearest street with the same > name? > > -- > Eric Ladner
_______________________________________________ Imports mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports
