[MetaCRS] "FOSS Maintained" Source of CRS Definitions

[Mateusz Loskot]

Landon,

I'm just an observer and CRS cadet who want to learn more about the
subject here, but I'd have a few, rather obvious comments.

Landon Blake wrote:
> I was wondering if there had ever been any effort to consolidate these
> different sources of definitions into a single data product maintained
> by an "open source" organization like the OSGeo. Maybe this data product
> is an XML file, maybe it is a group of text files, or maybe it is
> something else that I haven't thought of.

IMHO, the first thing would be to define common data model for all types 
of definitions and their sources. XML, text files, db tables, etc. are 
just physical representation of a model. If there would be a data model, 
then, I believe, it was possible to use any type of storage you can 
think of.

> I suspect each of the different CRS programming libraries has someone
> that is trying to keep up with changes and additions to the CRS
> definitions from the different sources mentioned above. I wonder if
> working together to consolidate multiple CRS definition sources into a
> single "FOSS maintained" source might save us all some work.

Yes, there are tools and scripts used in GDAL/OGR and PROJ.4 to update.
I think other CRS projects have their own tools.

> There is already a "standard" and open format for CRS definitions,
> namely WKT, correct? Couldn't this be used? (I know there are thousands
> of CRS definitions, but we could use streaming access to a text file, at
> least on the Java side of things.)

Perhaps it's a reasonable option. I'm sure there would be need of number 
of functions morphing between WKT and various formats (see 
OGRSpatialReference::morphToESRI/morphFromESRI). We've already 
identified in GDAL/OGR that we need another pair of morphers: 
morphToOracle/morphFromOracle. It indicates we have to deal with WKT 
variations.

> At any rate, maybe there is already some type of "single source" for CRS
> definitions already in existence. But I thought it didn't hurt to ask.

AFAIK, there is no common source but I'm sure that a gang of inventors 
of something like that would be blessed :-)

> It would seem that this would be one area that programmers using
> different languages could cooperate.

I'd list:
- CRS model spec.
- CRS interface spec. (not necesary but would be nice)
- common set of CRS data generators from X known resources to single 
repository
- common rules for CRS def. translators, so all import/export operations 
always result in the same output, independently to implementation.

Greetings
-- 
Mateusz Loskot
http://mateusz.loskot.net