[OSM-dev] likenesses

[Frederik Ramm]

Hi,

Eugene Alvin Villar wrote:
> Inasmuch as the underlying concepts are identical (more or less), I 
> don't like the idea that we can use different key name or values to 
> represent those concepts.

Thing is, the underlying concepts will very often be "more or less" 
identical but never really.

At the risk of expanding this into the vapourware realm, we have a very 
similar problem with "hierarchic tagging" which we sometimes try to 
solve by chain-tagging like in this fictional example:

natural=water
water=lake
lake=freshwater

We do this so that a renderer that only knows how to render water areas 
can work with this; and a renderer capable of distinguishing freshwater 
mountain reservoirs from quarry ponds (oops... not "natural", is it?) 
also has the necessary information.

Now if we introduce a likeness layer as SteveC suggested, that may also 
serve as a general "what is what" structure. The fact that a quarry pond 
is basically a lake and a lake is basically a body of water would no 
longer have to be explicitly tagged; it would be sufficient to tag 
"man_made=quarry_pond" and any renderer just interested in bodies of 
water could know that from the likeness layer.

This would make the likeness layer a central point of tagging, and free 
edit access to the likeness layer would become crucial for everyone, but 
at the same time it would allow anybody to invent the most crazy tags 
and say "oh well, if the renderer does not understand llwerewta=uwerwssc 
then it may treat this as an ordinary body of water".

Bye
Frederik