[GeoJSON] RFC-001 vote

Sean Gillies sgillies at frii.com
Wed May 23 07:45:21 PDT 2007


Christopher Schmidt wrote:
> On Wed, May 23, 2007 at 09:02:00AM -0400, Allan Doyle wrote:
>> Note that there is a counterproposal in http://lists.geojson.org/ 
>> pipermail/geojson-geojson.org/2007-April/000126.html
> 
> After working with GeoJSON long enough, I've realized that Tim is
> absolutely right on the money with this:
> 
> "Again, I've heard the argument that you can cram much of your parser 
> into one line if you keep the coordinates array identical for 5 of the 
> geometries.  Personally, I think this argument is a bit weak - and makes 
> the data structure overly funky."
> 
> I'm entirely in favor of:
> 
>  * Getting rid of 'holes' (polygons are a list of rings, first one is
>    outer) 
>  * Points being a tuple ([x,y])
>  * Lines being a list of tuples [[x,y], [x2,y2]]
>  * Polygons being a list of linear rings:
>    [[[x,y], [x2,y2], [x3,y3], [x,y]],[[x4,y4], [x5,y5], [x6,y6], [x4,y4]]]
> 
> I know I made the counter argument to begin with. It was stupid of me. I
> learned better as I wrote more code. 
> 
> So, I vote for doing whatever needs to be done to make Tim's suggestion [1] 
> the way things are done. If someone tells me the right way to proceed,
> I'll turn what he wrote into a proposal on the wiki. 
> 
> [1] http://lists.geojson.org/pipermail/geojson-geojson.org/2007-April/000126.html
> 
> 
> Regards,

Nothing wrong with a change of heart, particularly if it's a change 
based on implementation experience. I'd be satisfied with the coordinate 
representations above.

Sean




More information about the GeoJSON mailing list