Hi Jody,

Thanks for  your response.

I agree about the head hurting! Mine does already after only just brushing the surface!

Unfortunately the exact same problem exists with WKT. In order to make a valid WKT string for NZTM I would have to make a special case to put the y axis first and the x axis second. Otherwise it is just exactly the same as constructing geometries directly.

Now that I have done some more investigation trying out WKT as you suggested, I am pretty certain this can be considered a bug in ReferencedEnvelope; consider:


    //NZMG is an x/y CRS
    CoordinateReferenceSystem nzmg = CRS.decode("EPSG:27200");
    //NZTM is a y/x CRS
    CoordinateReferenceSystem nztm = CRS.decode("EPSG:2193");

    Envelope env = new Envelope(2345678, 2345680, 6543210, 6543212);
    ReferencedEnvelope refenv = ReferencedEnvelope.create(env, nzmg);
    System.out.println("NZMG Envelope: " + refenv);
    System.out.println("getMinimum(0): " + refenv.getMinimum(0));
    System.out.println("getMinX(): " + refenv.getMinX());
    refenv = refenv.transform(nztm, false);
    System.out.println("NZTM Envelope: " + refenv);
    System.out.println("getMinimum(0): " + refenv.getMinimum(0));
    System.out.println("getMinX(): " + refenv.getMinX());


which outputs

NZMG Envelope: ReferencedEnvelope[2345678.0 : 2345680.0, 6543210.0 : 6543212.0]
getMinimum(0): 2345678.0
getMinX(): 2345678.0
NZTM Envelope: ReferencedEnvelope[5980771.101482246 : 5980773.102397159, 1435407.1134907207 : 1435409.1144056704]
getMinimum(0): 5980771.101482246
getMinX(): 5980771.101482246

which is all correct, EXCEPT the last line. The minimum X is NOT the minimum of axis zero, and as ReferencedEnvelope is aware of the CRS and in all other cases the CRS is aware that it is y/x should not ReferencedEnvelope also be aware?

Were this to be fixed, might I suggest that then adding setMinX (etc) methods to ReferencedEnvelope which were likewise CRS aware would be a small step and (conveniently) would also resolve my original situation!

Thoughts? Thanks!
Iain




On 10/12/2016 11:25 a.m., Jody Garnett wrote:
Wow that NZTM situation is a mess; getting the wrong definition into the EPSG database takes some skill.

Can you construct the WKT representation of what you need (to match your data as provided) and then use GeoTools to reproject to your final answer? I have found that making special cases based on axis order makes my head hurt, and if I set up my definitions correct GeoTools can construct the required transform ...

--
Jody Garnett

On 6 December 2016 at 17:44, Iain Matcham <[email protected] <mailto:[email protected]>> wrote:

    Hi all,

    I have an application where I want a user to be able to enter a set of
    bounds and a CRS and then the code will reproject their input into a
    default CRS for further processing.  A fairly simple operation:

       public ReferencedEnvelope getEnvelope(
             double minEast, double maxEast,
             double minNorth, double maxNorth,
             CoordinateReferenceSystem crs) throws TransformException,
    FactoryException {
         Envelope envelope = new Envelope(minEast, maxEast, minNorth,
    maxNorth);
         ReferencedEnvelope refEnv =
    ReferencedEnvelope.create(envelope, crs);

         return refEnv.transform(DEFAULT_CRS, true);
       }

    Except for NZTM.  Thanks to LINZ, NZTM is has an axis order y/x. The
    above code falls over.

    Following this (admittedly quite old) thread:
    
http://osgeo-org.1560.x6.nabble.com/Need-help-reprojecting-ReferencedEnvelope-td5069254.html
    
<http://osgeo-org.1560.x6.nabble.com/Need-help-reprojecting-ReferencedEnvelope-td5069254.html>
    to the suggested resolution seems to be to "Make a special case for
    NZTM".  Extending this to "make a special case for any y/x CRS"
    means my
    code now looks like:

       public ReferencedEnvelope getEnvelope(
             double minEast, double maxEast,
             double minNorth, double maxNorth,
             CoordinateReferenceSystem crs) throws TransformException,
    FactoryException {

         AxisDirection axisDir =
    crs.getCoordinateSystem().getAxis(0).getDirection();
         Envelope envelope = axisDir == AxisDirection.NORTH ?
             new Envelope(minNorth, maxNorth, minEast, maxEast) :
             new Envelope(minEast, maxEast, minNorth, maxNorth);

         ReferencedEnvelope refEnv =
    ReferencedEnvelope.create(envelope, crs);

         return refEnv.transform(DEFAULT_CRS, true);
       }

    which seems like a lot of faff for what should be an easy operation.
    But we haven't even started yet because I am pretty sure this code is
    incomplete.  I _think_ the real question is not are we x/y or y/x
    it is
    are we right handed (x/y) or left handed (y/x).  So my code above
    would
    actually have to handle not just NORTH/EAST (right) and EAST/NORTH
    (left) but also NORTH/WEST (right) or NORTHEAST/SOUTHEAST (left)
    etc etc
    and so on through all the permutations of AxisDirection.


    So ... the only sensible resolution I can see without writing an
    entire
    novel of if/elses for all the possible pairs of AxisDirections is
    to use
    forceXY, but referring back to the discussion mentioned above, Ben
    strongly discourages this.

    Does anyone have any suggestions how this can be handled in a generic
    non-complex way without forcing x/y axis ordering? Hopefully
    someone can
    help.

    If not can I suggest a method

    ReferencedEnvelope.create(Envelope env, CoordinateReferenceSystem crs,
    boolean envelopeIsGeometric)

    which (if arg2 is true) treats the envelope as x/y and does any
    necessary axis conversion internally so that this piece of code can be
    written once and doesn't need to be written by anyone and everyone who
    might need NZTM in their app?

    Would anyone else find this useful?

    Thanks
    Iain

    --
    Sent from my ZX80


    ---
    This email has been checked for viruses by Avast antivirus software.
    https://www.avast.com/antivirus <https://www.avast.com/antivirus>


    
------------------------------------------------------------------------------
    Developer Access Program for Intel Xeon Phi Processors
    Access to Intel Xeon Phi processor-based developer platforms.
    With one year of Intel Parallel Studio XE.
    Training and support from Colfax.
    Order your platform today.http://sdm.link/xeonphi
    _______________________________________________
    GeoTools-GT2-Users mailing list
    [email protected]
    <mailto:[email protected]>
    https://lists.sourceforge.net/lists/listinfo/geotools-gt2-users
    <https://lists.sourceforge.net/lists/listinfo/geotools-gt2-users>



--
Sent from my ZX80



---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most 
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
GeoTools-GT2-Users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-gt2-users

Reply via email to