[mapserver-dev] bug in WFS writer?

[Bart van den Eijnden (OSGIS)]

Steve,

you were absolutely right, it seems to be a client problem. Both 
TatukGIS and ArcGIS make mistakes with these geometries, and Snowflake 
and Gaia3 are the one that do it correctly. Although there is a slight 
problem with Gaia3 for geometry no. 7.

See:

ArcGIS:
http://www.osgis.nl/download/umn/arcgis_kavels_gml.gif

Snowflake:
http://www.osgis.nl/download/umn/snowflake_kavels_gml.gif

Gaia3:
http://www.osgis.nl/download/umn/gaia3_kavels_gml.png

Sorry for the confusion!

Best regards,
Bart

Bart van den Eijnden (OSGIS) wrote:
> Okay, I found it already:
>
> http://trac.osgeo.org/mapserver/changeset/7164
>
> Best regards,
> Bart
>
> Bart van den Eijnden (OSGIS) wrote:
>> Hey Steve,
>>
>> Do you remember the bug number or changeset for that point-in-polygon 
>> fix? It sounds like it could be the culprit indeed.
>>
>> You're right it could be a problem in TatukGIS theoretically, but in 
>> this case things were checked also in ArcGIS. So it would be strange 
>> for 2 clients to have it wrong, though still possible.
>>
>> Mapserver version: 4.10.2.
>>
>> I haven't got the source data as yet (it's in Oracle Spatial), but 
>> I'll try to get it.
>>
>> Thanks.
>>
>> Best regards,
>> Bart
>>
>> Steve Lime wrote:
>>> Couple of questions:
>>>
>>>   - how are you defining a "problem", based on TatukGIS's rendering 
>>> or some other method? I'm not saying there isn't a problem but how 
>>> do you know it's not a problem with the renderer?
>>>   - do you have the source data (a shapefile?) available for testing?
>>>   - what version of MapServer?
>>>
>>> I recall another bug recently similar to this. It had to do with the 
>>> MapServer point-in-polygon being indeterminate when a point in one 
>>> part is on a polygon edge. The function is key to determining inner 
>>> and outer rings. I thought I fixed it though, at least in SVN. Since 
>>> a legal polygon can only share 1 vertex with an edge I updated the 
>>> code to check at least 2 and one more to break the tie if one 
>>> happened to be on an edge. That worked in any testing I had done.
>>>
>>> Steve
>>>
>>>  
>>>>>> On 3/19/2008 at 1:16 PM, in message <47E15866.2010407 at osgis.nl>, 
>>>>>> "Bart van den
>>>>>>         
>>> Eijnden (OSGIS)" <bartvde at osgis.nl> wrote:
>>>  
>>>> Hi list,
>>>>
>>>> it seems Mapserver WFS has a problem with geometries having a 
>>>> (common) point in them which touches the geometry itself. Mapserver 
>>>> WMS seems to draw these polygons okay.
>>>>
>>>> An example (look for instance at the second and third geometry):
>>>>
>>>> WMS:
>>>> http://www.osgis.nl/download/umn/wms.png
>>>> WFS in TatukGIS (screendump):
>>>> http://www.osgis.nl/download/umn/tatuk.png
>>>> WFS xml file:
>>>> http://www.osgis.nl/download/umn/wfs.xml
>>>> Any ideas what could be causing this? Does this ring a bell to anyone?
>>>>
>>>> Best regards,
>>>> Bart
>>>>     
>>>
>>>
>>>
>>>   
>>
>>
>
>


-- 
Bart van den Eijnden
OSGIS, Open Source GIS
bartvde at osgis.nl
http://www.osgis.nl