|
I try to map a disrupted postcode area to an area that also matches landuse="residential". I thought I'd use the German approach with multipolygon ... The village lies in Luxemburg near the German border. I couldn't find any advice on how the Luxemburgish mapping community addresses postal_code, but Nomanitim doesn't return any result for Luxemburgish Postcodes so far, although the mappers applied postal_code to streets directly (never saw that before). So I think it should be solved differently. I try to describe it in a very simple way: suppose we have 2 postcodes A and B in village X. They are "stacked" like this, with A being the old center of the village BAB prefectly matches the landuse="residential" area.
Is it OK to have at the end 3 mutlipolygons sharing the common ways, all components specified as "outer"?
Thanks for your advice, Georges |
|
Yes, you solution sounds good! There are a bunch of such separated areas with the same postal_code in Germany, too. If you have done your edit, give us a permalink to that town, then we can inspect the relations. There are also some ways to visualize boundary areas mapped as relations, there you can even detect logical errors in boundary elements. More to come ... I was almost sure that my approach must be OK, so I uploaded the data already yesterday. And Nomanitim shows exactly the result I was expecting (hoping) to see with "lenningen 5431". "5430" is the other postcode. http://www.openstreetmap.org/?relation=3163883#map=16/49.6011/6.3658 But to which extend is this applicable? I had already to guide one border through connected buildings (left near the church). But there is apparently another building in the north of the village lying in a second row of postcode 5431 but belonging to 5430. Should I really draw an island area with role=inner for multipolygon 5431 and role=outer for multiploygon 5430. Isn't that overkill for one building? In Luxemburg you have 4 digits postcodes with round about 500 localities. You can imagine that often a postcode only covers a couple of streets. Sometimes you can barely call them areas. If you have just 2-5 streets covered by a postcode, a simple postcode relation tying these streets together would be much simpler. But I'm afraid of the many different algorithms client software have to provide to analyze all the concurrent strategies to organize hierarchical data.
(26 Aug '13, 21:55)
gjesch
Hmmm, comparing to postal_code system in Germany, maybe it is too easy to draw a boundary around a bunch of streets and houses, and to define by this "little" area that this the area with postal_code=xxxx Only for a general compare: is there an official map online where we can see the system of postal_code areas in Luxembourg? If you can speak German a little bit, then I invite you to the german section at http://forum.openstreetmap.org ... there we have a quite recent topic about the correctness of postal_code areas in Germany.
(28 Aug '13, 16:27)
stephan75
|
