[GeoJSON] Current state of the RFC

[Sean Gillies]

Allan Doyle wrote:
> I was just going to "clean up" the RFC, thinking it would be a simple  
> matter of dotting some I's and crossing some T's...
> 
> Not quite... we now have two distinct classes of geometry types.
> 
> 1. Those which have "geometry" in the feature
> 
> 2. Those which have "members" in the feature
> 
> This means that you can no longer use the existence of geometry.type  
> as a means of figuring out what you're dealing with.
> 
> Furthermore, you can't a priori tell the difference between a  
> MultiLineString and a GeometryCollection or a MultiPolygon and a  
> GeometryCollection.
> 
>  From my personal perspective, this does not bother me, I don't feel  
> I would ever need the multi's or the geometry collections, but for  
> those of you who do, I think it warrants a little more thought.
> 
> One solution would be to simply drop all the multi-kruft and stick  
> with Point, LineString, Polygon, and Box. I'm inclined to do that and  
> let the multi-people write RFC-002, but I can be swayed by cogent  
> arguments.
> 
> 	Allan
> 

We could reconcile geometry types by expressing multis in a feature like

{ ...,
   "geometry": {
     "type": "MultiPoint",
     "members": [
       { "type": "Point", "coordinates": [[...]] },
       ...
     ]
   }
}

Multi types would have a members object, an array of sub-geometries.

-- 
Sean Gillies
http://zcologia.com/news