[OSM-talk] River Avon in Mapnik

[Jon Burgess]

On Tue, 2007-10-23 at 13:41 +0200, Dirk-Lüder Kreie wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> David Groom schrieb:
> >> ----- Original Message ----- 
> >> From: "David Groom" <reviews at pacific-rim.net>
> >> To: "mike" <mike at bristolbeat.co.uk>; <talk at openstreetmap.org>
> >> Sent: Tuesday, October 23, 2007 11:11 AM
> >> Subject: Re: [OSM-talk] River Avon in Mapnik
> >>
> >>
> >>
> >> It depends on how you see things but it could be both a tagging and a 
> >> rendering problem.
> >>
> >> The riverbank was originally derived from the PGS coastline import script, 
> >> and the data then tidied up by me.
> >>
> >> The import script tagged the riverbanks as natural=coastline, and I left 
> >> it as such for two reasons:
> >>
> >> 1) no clear definition of where a coastline ends and a river begins when 
> >> it is a tidal river.
> >>
> >> 2) In Osmarender the rendering of coastline areas in my opinion is a lot 
> >> less problematic than rendering of river areas, with a coastline all you 
> >> have to do is ensure the "segments" point the right way, with water on the 
> >> right,
> > 
> > I should have added a point (3)
> > 
> > 3)  To get an area waterway = riverbank rendered in mapnik it has to be a 
> > closed area. This means drawing a "segment" across the river to join each 
> > side of the riverbank.  I know this will render correctly, but it seems 
> > "wrong" to me to mark a riverbank across the river when no such feature 
> > exists in real life.  I guess it depends on whether you see the data we are 
> > producing as being tagged to draw pretty maps in mapnik, or to actually 
> > represent what exists in real life.
> > 
> > As I said earlier, I rather hope relations might solve the problem.
> 
> This should already work with a multipolygon relation, but I'm not sure
> what osmarender does with the missing ways.

Do multipolygons support this?
I understood that the multipolygon relation was for defining polygons
with holes and not for handling large polygons split into multiple
ways. 

	Jon