[Talk-GB] Relations

Lennard ldp at xs4all.nl
Tue Mar 2 21:19:39 GMT 2010


Ed Loach wrote:

> I think there are only two countries in the world (according to the

Three.

> wiki) who strangely think it better to use multipolygon relations
> rather than boundary ones. For those of us who edit (and correct)

That is because boundary relations at their core *are* multipolygons 
(the OSM kind). They are built up exactly the same, with outer rings and 
possible inner rings. It's just the terminology that's different.

And at the moment there's two of these that tools have to handle the 
same, but what if someone invents a foobarbaz relation that also works 
with these rings? Should all tools then add support for the foobarbaz 
relation, which internally works exactly the same as a multipolygon 
relation? Instead of doing type=multipolygon + foobarbaz=yes?

Why do we need a superfluous type=boundary + boundary=administrative 
when type=multipolygon + boundary=administrative works exactly the same? 
The defining quality of a boundary relation isn't the type. It's the 
boundary=administrative bit. The type=multipolygon then only triggers a 
unique handling of this geometry in all tools that support them.

> boundary relations in Potlatch the highlighting of boundary relation
> members in purple rather than blue for multipolygons makes things so
> much easier (especially at the moment in GA where I'm finding
> landuse multipolygon imports by county butting up against the
> boundary way).
> 
> So yes, I've considered it, and consider boundary worthy of being a
> special case of multipolygon.

But you seem to have missed the point where you considered asking Mr. 
Author Man of your $fav_editor to trigger the purple rendering based on 
boundary=administrative only/as well. Which to me makes all the more 
sense, instead of this whole "tagging for the editor" business.

-- 
Lennard




More information about the Talk-GB mailing list