[mkgmap-dev] --add-pois-to-areas behaviour

Thu Jan 28 15:42:18 GMT 2010

Felix Hartmann schrieb:

> Well you might not like to have a POI for every polygon (what's the
> point in a POI for a city? - cities are entered as seperate points anyhow)

Ok, I see there are problems. But I thought that there is an algorithm to check, if there exist a point in the polygon, yet to avoid double points for the same thing.

> There is absolutely no reason why a POI should be in the middle of a
> Polygon.

It is easy and a good approach.

> It would be stupid to say because we have a city boundary, we
> don't need a city POI in Openstreetmap. For a building you might want to
> have the POI at the entry, and not in the middle of the buidling, .....
> It is impossible to know whether if there exists a POI inside a Polygon
> (it will be never checked, cause too much cpu cycles to do so) it should
> be printed nevertheless, or it shouldn't be printed on the map.

But I thought there is exactly such algorithm integrated in mkgmap. Did I've seen ghosts?

But you are right, this would cost cpu cycles to do it. So I would suggest to provide an extra-option to enable such behaviour.

> There are many many more cases where it is stupid to assume that one can
> simply assign a POI to a Polygon.

Ok, maybe it would be good to define it separately for each feature.

I could imagine something like:

polygons:
building=yes {makepoint} [0x13 resolution 20]
# or just:
building=yes {makepoint}
# if you need only a point but no polygon!

points:
building=yes [0x6402 resolution 24]

Then you should be able to give an extra-option like:
--add-pois-to-areas=check-existing-points

so no points where created for a polygon if there is a real OSM-point that matches the rule (in our example building=yes)

Does it sound sensefull?

Christoph

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: OpenPGP digital signature
Url : http://lists.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20100128/17e5cbdc/attachment.bin 