[MetaCRS] RE: Initial commit of CSV Test data files

Martin Davis mbdavis at refractions.net
Mon Nov 30 15:12:23 EST 2009


Frank,

I can see your point about trying to be in synch with the EPGS specs 
(and thus make PROJ4 sensitive to axis order).  How are you planning to 
capture this information in the PROJ4 CS catalog?   Is there a parameter 
that will be added to PROJ4 CRS defs to indicate this?  Or is it just 
assumed that any geodetic coordinate is expressed using lat/long order?


Martin

Frank Warmerdam wrote:
> Norm Olsen wrote:
>> Interesting point.  EPSG does specify latitude, longitude, and then 
>> height.  My experience says that many (if not most) coordinate system 
>> libraries use the longitude, latitude, height convention.  That is, 
>> 99.8% of the time and regardless of whether the coordinate system is 
>> projective or geographic, the first ordinate increases to the east, 
>> the second ordinate increases to the north, and the third ordinate 
>> increases away from the center of the earth.  I don't know what the 
>> default preference for Proj4 is, so I'd be interested in knowing of 
>> the opinions of the Proj4 folk on this.
>>
>> Dogmatically, EPSG is correct.  Pragmatically, I believe longitude, 
>> latitude, and then height is correct. 
>
> Norm,
>
> PROJ.4 is still "axis orientation ignorant" so it does indeed assume
> long, lat, height for geographic coordinates.  However, the GDAL
> OGRSpatialReference and OGRCoordinateTransformation classes try to have
> some sort of knowledge of SRS axis ordering though this work is not 
> really
> complete at this time.
>
> While I hate the misery that the epsg lat/long axis police have caused
> in this world (mostly via more recent OGC specifications) I am not sure
> ignoring the EPSG axis definitions is the right thing to do.  I've got
> a foot on either side of this issue and so I've tried not to take a 
> position.
>
> In part I suppose it depends on whether our objective is for the test
> data to address broadly defined coordinate system transformations as
> opposed to being primarily focused on projection transformations.  If
> the former then we ought to expect test apps to honour the axis 
> definitions
> of the coordinate system. If the latter then sticking to a default axis
> orientation (perhaps only for geographic coordinates) would be fine.
>
> I can go either way.
>
> Best regards,

-- 
Martin Davis
Senior Technical Architect
Refractions Research, Inc.
(250) 383-3022



More information about the MetaCRS mailing list