Re: [Geotools-devel] Difference into two apparently equal CRS?

[Martin D. <mar...@ge...>]

Le jeudi 08 février 2007 à 09:55 +0100, Andrea Aime a écrit :
> I'd say, if we don't want to compare metadata, not even names should be
> considered, but only the ellipsoid parameters and the bursa-worlf 
> transform (provided the latter is available)?

No. Two datum with the same ellipsoid may be really different things.
The datum describes how the ellipsoid is "attached" to the earth. We may
have the same ellipsoid, but the ellipsoid may be slightly translated or
rotated (while having the same axis length) in one datum compared to an
other datum. The datum name is currently the only way to differenciate
two datum. This is not specific to Geotools, see:

http://gdal.velocet.ca/~warmerda/wktproblems.html

(The link doesn't seem to work for me right now, so I'm not completly
sure that it is the right link).

The workaround is to define a list of aliases. Geotools defines such a
list there:

http://svn.geotools.org/geotools/trunk/gt/modules/library/referencing/src/main/resources/org/geotools/referencing/factory/DatumAliasesTable.txt

For example the first line tell us that datum with name "Abidjan 1987",
Abidjan_1987" and "D_Abidjan_1987" are considered equivalent. If you
have the name of datum that should be considered equivalent, we could
add them to this list.

	Martin