[OSM-dev] Area data type -- again!

[Ilya Zverev]

Tobias Knerr wrote:
> >/  Malcolm Herring wrote:
> />>/  My other wish is that disjunct outers be dis-allowed. Where multiple
> />>/  complete areas need to be associated, then this is the proper use of
> />>/  relations.
> /
> +1
>
> >/  Why? How would you draw an administrative district with an exclave?
> /
> A relation for the administrative area with two area elements to
> represent the territory, plus possibly a member for the capital and
> whatever other members mappers come up with.
>
> As we have relations for representing the relationship between elements,
> I believe that we should design the area data type as a polygon (with
> holes), not as a multipolygon. After all, it would be odd to have a
> relationship between multiple areas as the one special case that works
> without relations, while every other combination of elements uses relations.

So... a park with two parts separated by a railway? A city with exclave 
(not administrative boundary, but place multipolygon now)? I live in 
such city now, and I have that kind of park nearby. One can come up with 
more examples, where multiple outer rings are needed.

Why should we restrict ourselves (adding extra tests to API by the way), 
just to have to circumvent this restriction later with relations? Why 
cannot administrative boundary relations be redefined to having an 
entire area as a member, instead of working as a crippled multipolygon 
for areas that were meant to replace multipolygon relations?

My opinion is that there is no much difference in having multiple versus 
one outer rings: the processing is still the same, but there would be 
less checks. And without a good reason to do that, I'm against 
restricting areas to simple polygons with a single outer ring.

IZ