Hi,
so I've tricked the cite tests and had GeoServer pass all of the
basic part (I still have a single failure in transaciton that I'll
handle later).
But unfortunately this is just a trick. I loaded into postgis the
data in lat/lon order (that is, the data the OGC provides, and that
does not respect the native Postgis axis ordering), and forced
GeoServer to use lon/lat axis ordering by adding the
-Dorg.geotools.referencing.forceXY=false option to the JVM.
This means that GeoServer is running assuming all data coming from
datastores is in lat/lon order, when we all know it's not (it is
in this particular case just because postgis was feeded with
data in that order).
What should we have to say to users, GeoServer is WFS 1.1 compliant
provided all your data sources are in lat/lon order? Sigh...
For datastores that do have a native notion of the SRS (postgis,
shapefile with a prj) would could allow the user to specify
the SRS as "urn:x-ogc:def:crs:EPSG:6.11.2:4326" and then have
GeoServer reproject on the fly from the native lon/lat to the
declared one.
But that would not work for a plain shapefile with no prj,
there we would have to add a "flip axis" to allow the user
stating the data is natively in lon/lat, but he wants to serve
it in flipped order, or else to allow the user to specify
what's the native WKT...
Oh, my poor head...
Cheers
Andrea