| Hi Jonathan, 
 Thanks for the clear images; those help illustrate the issue quite
    nicely!
 
 As some good news, I can explain the issue, hopefully provide a
    work-around (and a diagnosis step).
 
 With 'large' polygons, there is a corner case where it becomes
    challenging to know whether or not the desired polygon crosses the
    anti-meridian or not.  For your second case, the query re-writer
    uses Spatial4J and assumes that the connections between the upper
    left and right corners goes the other way around the planet.  That's
    why the result set seems to flip.
 
 Incidentally, for BBOX queries, there's no such confusion and hence
    those work just fine.
 
 The solution is to add additional way points so that successive
    coordinates have a difference in longitude less than 180 degrees.
 
 As a debugging step, check out the GeoMesa query explainer:
 http://www.geomesa.org/documentation/user/datastores/explaining.html
 http://www.geomesa.org/documentation/user/geoserver.html#geoserver-explain-query
 
 I think the filter re-writing would have already happened.  That'd
    give some insight into whether or not GeoMesa has flipped the query
    around.
 
 Cheers,
 
 Jim
 
 On 12/19/2017 05:09 PM, Jonathan Meyer wrote:
 
 
      
      
      
      
      
            I am using Geomesa with a  Cassandarda data
          store.  I have come across a potential issue that I’m not sure
          how to deal with and am hoping someone may be able to shed
          some light.  I have a front end web app using leaflet to
          display point data queried from Geoserver.  So I am using the
          Geomesa Cassandra plugin to geoserver.  I have an edge case
          where if a user enters a very large (80% of the world) polygon
          search area, the search results will return all the points
          outside the polygon and not inside the polygon.  Using a BBox
          instead of a polygon in the same approximate area will return
          all the points within the bbox as expected.  Drawing a polygon
          that is say 60% of the world will also return expected
          results.  There seems to be some point where polygon searches
          don’t return what is expected.   The only thing that changes
          in these searches is the shape or size of the geometry. Any help or suggestions greatly
          appreciated.   See the below images as examples:   
   
   Jonathan Meyer   
 
 _______________________________________________
geomesa-users mailing list
geomesa-users@xxxxxxxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.locationtech.org/mailman/listinfo/geomesa-users 
 |