SRID 3857 (instead of 3785) about rgeo HOT 3 CLOSED

Hi Greg,

Yes, this is a general source of headache. When implementing simple_mercator, I tried to bury most of the issues and develop a system that generally "just works" for the common cases:

• I chose to use SRID 3785, even though EPSG wants to deprecate it in favor of 3857. This was done because I can verify that 3785 is known by all spatial reference databases supported by RGeo (spatialreference.org, PostGIS, and Proj4). 3857 is missing from some databases.
• The actual transform implementation for simple_mercator uses custom (Ruby) code that performs the same actual transform that Google performs. It does not use the OGC parameters, precisely for the reasons given in the article you cite: it is not possible to express the actual Google transform using the OGC format. Therefore, the parameters are provided merely for informational purposes.

If we think the parameters are "wrong", we can make changes to them. Although I'm not sure it matters much, given that they aren't used by the factory to perform transforms, and given that, regardless of what we do to them, they'll be "wrong" anyway, in that they will never actually express the correct transform.

from rgeo.

I see. Well, maybe one for the docs then. I just wanted to verify that this wasn't going to shift by 20km some day when it was addressed.

from rgeo.

Is there a plan to shift over to EPSG:3857 at some point?

from rgeo.