[OSM-talk] cadastre, ways, rendering

[Kristian Thy]

On Tue, Jan 23, Marcus Bauer wrote:
> Until renderers have artificial intelligence, they will need some help
> to make decent maps. 
> 
> Have a look at this:
> http://www.openstreetmap.org/index.html?lat=43.67615439303823&lon=7.207250585123151&zoom=11

What about it? Looks pretty nice to me.

> To avoid this you either add more and more rules to the renderer, ad
> infinitum, or you help it a bit. You don't loose any information if only
> one side of a motorway has a ref=A5 key/value.

How does one side of the motorway know what is the other side then,
since the only thing that makes them reference eachother is, oh I don't
know, their ref value?
 
> If a highway has more than one ref, like A8/E88, it makes sense to
> alternatingly tag ways, i.e. one section of a highway A8, the next E88.
> You will always have more information to put into the map, but it
> doesn't get better and isn't even of use for later routing, as long as
> it is tagged motorway.

I'm speechless (almost).

You talk about changing data until renderers have artificial
intelligence, but the things you are changing it into will actually
remove information from the dataset. If the map is cluttered, select
less items to render - don't delete data from the base set.

> Another example: how do you make curves? Actually you add points until
> it nicly renders at higher zoom levels. So in a certain way you "change"
> data to suit the renderer.
> 
> The point is to find a good compromise. 

This whole discussion is about the attributes or metadata. It is clear
and understood that what is in reality a fluidly linear feature will be
represented in data as a discrete collection of vertices. Even so,
renderers could choose to interpolate spline or bezier curves between
nodes if desired.

\\kristian
-- 
Cthulhu.call();