Ethan:
The structure of a URN is so that you can create one - NOT so that they
are parsed for content. I think the fuzziness at the OGC in this regard
is simple mistaken.
GML is used now in the http://www.epsg-registry.org. This registry and
the assignment of EPSG codes is controlled by the registration
authority, in this case the OGP. Another such registry could easily be
deployed assuming that someone is willing to be the registration
authority and this could have perhaps a difference governance model than
used by the OGP. The 19111 new parts will make their way into GML of
course.
Ron
-----Original Message-----
From: Ethan Davis [mailto:edavis@xxxxxxxxxxxxxxxx]
Sent: October 9, 2008 12:15 PM
To: Ron Lake
Cc: Unidata GALEON
Subject: Re: [galeon] How to map a parameterized projection to an EPSG
code
Hi Ron,
I agree that a URN should only be an identifier. And one could argue
that since they are URNs that all OGC URNs really are only identifiers.
However, since the OGC CRS URN "naming" scheme is so detailed,
parameterizable, and even compoundable, I think that if one cared
(dared?) to parse the URNs one could figure out the definition of the
CRS. So, I think for OGC CRS URNs, the line between identify and define
is pretty fuzzy.
I also agree that it would be good to have a way to easily register
commonly used, parameterized CRS. And a grammar in which the CRS can be
defined. (Is GML up to this task? Or will it be once ISO 19111-1 and -2
are folded in?)
On the other hand, for less commonly used or somewhat unique CRS, making
the definition available at some URL and using that URL to reference the
CRS may be enough.
Ethan
