Am 21.11.2012 13:27, schrieb Thorsten Kukuk:

Hi,

with the current splitter/mkgmap, I have a flooded tile in norway. The reason seems to be multipolygon 96565, "Randsforden". The mulitpolygon is complete, but it is tagged as "natural=coastline". But it is inside of Norway, so no real coastline in my opinion.

I think this shouldn't be "natural=coastline" but "natural=water". My questions are only: - should this be changed? I think yes. Yes, this would be better. ;) - why does this flood the tile? Do you use precompiled coastlines?

Henning

On Wed, Nov 21, 2012 at 01:27:33PM +0100, Thorsten Kukuk wrote:

with the current splitter/mkgmap, I have a flooded tile in norway. The reason seems to be multipolygon 96565, "Randsforden". The mulitpolygon is complete, but it is tagged as "natural=coastline". But it is inside of Norway, so no real coastline in my opinion.

I think this shouldn't be "natural=coastline" but "natural=water". My questions are only: - should this be changed? I think yes.

FWIW, whenever I encounter a natural=coastline that should not be so, I will either add the polygon to a multipolygon relation and remove the natural=coastline tag (so that only the multipolygon relation will carry area-defining tags), or replace the natural=coastline with natural=water. I guess that some Potlatch users do not know the difference between water and coastline.

- why does this flood the tile?

The relation 96565 looks bogus to me. The name and natural=* tags should be on the relation, not on the outer way. If the natural=coastline is on the role=outer line (Randsfjorden (8758490)) then mkgmap is entitled to flood the tile. :-) Marko