[mapserver-dev] Axis order in WMS 1.3.0 (#3582)

[Yewondwossen Assefa]

On 03/05/2011 3:34 PM, Martin Kofahl wrote:
> Hi,
>
> in my mind MapServer still doesn't handle the axis order correctly for WMS 1.3.0 requests. The current implementation does first check a static list of epsg codes about axis orientation and otherwisely swap axes for epsg 4000-4999. (See http://trac.osgeo.org/mapserver/changeset/11155#file4, http://trac.osgeo.org/mapserver/browser/trunk/mapserver/mapaxisorder.csv?rev=11155)
>
> Epsg codes with swapped axes commonly used in Germany are currently neglected and there are probably many more. As more and more WMS 1.3.0 clients come up, users get confused about WMS suddenly deliver a blank map (because of misinterpreted bbox).
>
> MapServer makes the one who compiles MapServer responsible for the swapped-axis-codelist (mapaxisorder.csv). That's why I worry that there will be different MapServer 6 versions running with different handling of the axes order for identical requests.
>
  I think this last paragraph is somehow incorrect, MapServer uses 
mapaxisorder.h to compile.  mapaxisorder.csv and the script 
mapaxisorder.sh are mechanism to generate mapaxisorder.h. This file is 
committed. So if someone compiles out of the box, there is no reason to 
think that he would get a different result.
The mechanism is there so that it is easy for people to generate 
mapaxisorder.h instead of editing it by hand. If they want, they can 
then contribute back their epsg codes and we will generate a new 
mapaxisorder.h and commit so others would have them.
The initial list that I put there is based on the epsg listed in the bug 
as being "inversed" coordinates but not working currently with the logic 
of assuming that all espg codes between 4000-5000 are "inversed".
> The best would be to make MapServer axes-orientation-aware. Using the sql snippet from http://trac.osgeo.org/mapserver/ticket/3582#comment:13 I got about 1700 epsg codes with north/east (resp. south/west) axes, see http://trac.osgeo.org/mapserver/attachment/ticket/3582/mapaxisorder.h. I'm no expert in foreign epsg codes. From my understanding they all should be marked having swapped axes when processing WMS 1.3.0 requests. Someone who con confirm this? What about implementing the whole list into MapServer ... and hoping the best?
>
  I am not sure what is the best way to know all the epsg code that are 
inversed. I was more under the impression that this list in 
mapaxisorder.h would be more of an  an exception list than a long list 
of >1000 elements. Are you sure about this list provided in the bug? I  
am also a bit suspicious about it since I do not see 4326 which I know 
should be inverted.

best regards,

> Best regards,
> Martin
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>


-- 
----------------------------------------------------------------
Assefa Yewondwossen
Software Analyst

Email: yassefa at dmsolutions.ca
http://www.dmsolutions.ca/

Phone: (613) 565-5056 (ext 14)
Fax:   (613) 565-0925
----------------------------------------------------------------