[OSM-talk] House numbers... One more suggestion

Karl Newman siliconfiend at gmail.com
Mon Jul 28 18:59:28 BST 2008


On Mon, Jul 28, 2008 at 10:21 AM, Charlie Echo
<openstreetmap at coutiere.com>wrote:

> If things are clear, and if there is a consensus about using this Karlsruhe
> schema, let's have a vote on it.
>
> This will make things easier: we would then have a clean situation. This
> would enable people to enter the data.
> We may also adapt the tools then, e.g. create a function in JOSM that
> "explodes" a way by creating parallel ways (for each continuous segment)
> with pre-filled tags (street names, and so on) so that entering data is easy
> enough.
>
> Charlie Echo
>
>
The Karlsruhe schema is fine for just showing numbers on a map, but I don't
like it because it's not topological and therefore virtually impossible to
use for other purposes. This is because the associated way is not directly
encoded in the schema--it has to be derived by another means. That's just a
whole lot more work for data consumers that are only trying to locate a
street address a relative distance along a way (e.g., all current GPS
navigation systems). Consider this use case: Given a street, I'd have to
look through NN million nodes to find the closest ones (say within some
radius, which may actually miss some!), and then for each of those nodes,
check against MM million ways to make sure they aren't actually closer to
another street. That's madness! A simple relation could tell you node X in
way Y has the address number Z. (Obviously need a little more detail such as
odd/even schemes to fully support interpolation).

Karl
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20080728/9fbee822/attachment.html>


More information about the talk mailing list