[Geojson] polygon point order

Christopher Schmidt crschmidt at metacarta.com
Mon Oct 15 06:51:34 PDT 2007


On Mon, Oct 15, 2007 at 02:41:35PM +0100, Martin Daly wrote:
> > Two possibilities mentioned so far:
> > 
> > A. "Law of the minimum" =
> > > always take the smaller interpretation (the one < 180 degrees wide).
> In
> > > exchange for that maximum size interpretation, you can ignore
> ordering,
> > > and the rule works for any line, not just closed polygons, the way
> > > clockwise/anti-clockwise does.
> > 
> > B. "Clockwise, west-to-east"
> > 
> > Simplicity seems to favor A.  Are there any cases where A would fail?
> > (If a user really wants a 181-degree line, then maybe the client could
> > insert a vertex mid-way.)
> 
> Or C. "Law of the spheroid"
> 
> Allow coordinates/angles <-180 and >180.

This is what OpenLayers does, for the most part. (With the non-default 
date-line wrapping, it does A, I think, if I understand, but vectors
are not well supported in that case.)  It also means you're not special
casing your geometry description language for a specific
projection (unprojected lat/lon).

> Or D. "Law of the jungle"
> 
> Do nothing.
> 
> I like D.  I'm queasy about being the first to come up with a rule to
> follow.

Also throwing in my hat for D, in case it wasn't clear from what I said
earlier :) Though Since C implies "Don't add any limits on the bounding
box for features" -- and there currently isn't one in the spec -- I
think that D/C are both 'current state'. 

Someone else said at one point "If you're worrying about date-line
wrapping, you're using the wrong projection." I don't know if I 100%
agree, but I think that it's something to consider.

Regards,
-- 
Christopher Schmidt
MetaCarta



More information about the GeoJSON mailing list