[OSM-talk] shp-to-osm 0.7

[Frederik Ramm]

Hi,

Pieren wrote:
> Apart from this tool, I don't like the idea of always putting the tags
> in the relation. If you only have tags on the outer ring, why not. But
> we also have very often tags on the inner ring. And I don't see why we
> should have sometimes the tags in the ways and sometimes in the
> relations just because they use a special relation type. I could also
> say that the tags in the inner ring have the same right to be in the
> relation as the outer ring.

Having tags on the inner ring is a compromise between correct modeling 
and saving work for the mapper.

In computational geometry, a multipolygon can have holes, but these 
holes do never have an identity of their own - they are just the absence 
of whatever the multipolygon represents. Thus a hole in a forest is 
simply something where there is no forest - it doesn't say what is there.

If you wanted to model things correctly then you would have one area 
object for the multipolygon (e.g. the forest), and another area object 
for the area inside the hole (say a lake).

multipolygon #1
   member role="outer" way=#10
   member role="inner" way=#11
   tag natural=wood

multipolygon #2
   member role="outer" way=#11
   tag natural=water

It's just that we do not want to create so much work for the mappers, 
that's why we allow a shortcut: "If you have an area that has no holes 
and just one outer ring, then you can tag the outer ring accordingly and 
it will be recognized as a polygon".

So yes, I do agree with you that this is not "clean". But I do not agree 
with your conclusion ("just because we allow some things to not be 
clean, let's just drop clean modelling altogehter").

Bye
Frederik