Re: [Geotools-devel] problems with coordinate system flipping

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

Justin Deoliveira a écrit :
> The problem that I am running into is that in GeoServer we use the hint
> the to switch the axis of 4326. However when a WFS request comes in with
> the srsName of:
> 
> "urn:x-ogc:def:crs:EPSG:6.11.2:4326"
> 
> The hint does not seem to apply and the coordinate system axis are not
> swapped. This is a problem because it retults in a conversion between
> the two and all geometry x,y values are swapped.

This is http://jira.codehaus.org/browse/GEOT-1388

Basically the current behavior is intentional. We could change it but I don't
know what should be the correct behavior. I though that the whole point about
"urn:ogc:..." syntax was to fix the confusion around "EPSG:xxxx" syntax. OGC has
been very clear that the policy for non-legacy specifications is "as the
authority said, no axis flip". This is especially clear in WMS 1.3.0:

   "EPSG geographic coordinate reference systems follow ISO 6709 and
    always list latitude before longitude." (page 9)

   "EXAMPLE: EPSG:4326 refers to WGS 84 geographic latitude, then longitude.
    That is, in this CRS the X axis corresponds to latitude, and the Y axis
    to longitude." (page 10)

The problem is what to do with legacy applications and specifications like WMS
1.1. We have to live with the fact that "EPSG:4326" in WMS 1.1 is interpreted as
(longitude,latitude). But "urn:ogc:...:epsg:4326" has been introduced later, and
I though that it was supposed to be a "fixed" syntax.

Andrea and I wrote an email to Arliss Whiteside, who is the author of OGC 06-023
specification which defines the "urn:ogc..." syntax. Arliss told us:

   "Yes, WMS 1.1 and 1.3 specify opposite axis orders for "EPSG:4326",
    as Carl Reed confirmed.  Furthermore, neither specifically allows
    use of URNs to reference CRSs, such as "urn:ogc:def:crs:EPSG:6.3:4326".
    (...snip...) Perhaps worse, most other OGC specifications do not indicate
    how to reference CRSs.  Many existing implementations of these other
    specifications use the incorrect specification from WMS 1.1.  Apparently
    the CITE tests for WFS 1.1 include tests for this incorrect WMS 1.1
    meaning of "EPSG:4326", applied when the URN is used (which is more
    clearly incorrect)."

So my understanding of the last sentence is that applying axis flip on
"urn:ogc..." is incorrect, and the CITE tests may need to be fixed in this
regards - if "urn:ogc..." is widely interpreted in the flipped sense, we may
need to adapt to this reality. Andrea was supposed to contact the CITE guys in
order to ask them. I don't know what the result was.

So current behavior is intentional, I can change it but would like to know what
I'm supposed to do...

	Martin