[GeoJSON] Point as list of one point, or list of coords

Christopher Schmidt crschmidt at metacarta.com
Fri Apr 13 12:27:15 PDT 2007


On Fri, Apr 13, 2007 at 05:16:45PM +0100, Martin Daly wrote:
> > Good enough for me. I tweaked RFC-001.
> 
> I've further tweaked it, to add multi-s, polygon exterior and hole-s and
> (lovingly hand-crafted and therefore prone to syntax errors) examples of
> all of the geometry types.

My previous response was not really clear on how I feel.

 * I am in favor of a full JSON-based feature representation. (This
   obviously has to include multi-geometries.)
 * I am in favor of a 'simple' GeoJSON conformance level that does *not*
   require support for multiple geometries.

I've implemented a GeoJSON RESTful server. It maps JSON into KML,
GeoRSS, GML/WFS, and the OSM file format. It doesn't have multi
geometries. It lets me scribble on a map in OpenLayers and save the JSON 
data to the server.

It does pretty well. It works. And it has no support for multigeoms.

This is not a 'full' mapping of every possible type of data: it's a
description of data along the lines of what Google's "My Maps" service
can handle.

I think this level of functionality is extremely useful -- it's
translatable to many formats without the need for the full support of
GML and multiple geometries.

So, I feel that:

   There should be a 'simple' conformance which does not support 
   Multiple Geometries. This simple conformance only requires support
   for Point, Line, Polygon, Box. (I'm unconvinced on whether polygons
   need to support 'holes'/multiple rings.)

Then, I think a second level of conformance should be created, which
includes the functionality just described in the RFC-001 wiki page.

IN other words: I'd like RFC-001 to go back to what it was, and RFC-002
to be created with what RFC-001 currently is.
   
Regards,
-- 
Christopher Schmidt
MetaCarta



More information about the GeoJSON mailing list