[GeoJSON] Plug-Pull request to finalize a CRS-free I-D revision

[Howard Butler]

On Jun 14, 2013, at 1:36 PM, Tim Schaub <tschaub at opengeo.org> wrote:

> To clarify, I'm ok leaving it in as long as we don't touch the coordinate order.  The coordinate order we have now works well for clients interested in rendering (atop images that come with notions of top and left).
> 
> My primary motivation for wanting to remove the CRS member was because I know as long as we have it in there, people will argue that our coordinate order is wrong.  A secondary motivation is that I think all GeoJSON in one CRS is better for interoperability (I still believe that no matter how well you specify the CRS member, very few if any browser clients will be able to handle arbitrary CRS).

I agree with this and think language about both items should be put in an Errata section to give a little bit of background. 

> So while I think we're setting ourselves up for less interoperability in favor of efficiency, I won't object to doing nothing with the spec regarding CRS.

I think we're setting ourselves up for less interoperability in favor of deference to those who've already committed to the current specification. An example of guiding support of this thinking is GDAL. There are API decisions in GDAL that are dead wrong, yet cannot be changed because changes would break all kinds of client code. We suffer our sins and it makes us think (a little) harder next time. GeoJSON is only going to be around for a (software) generation or so.

Howard