Hi Roald, On Mon, Jul 18, 2011 at 2:36 AM, Roald de Wit <[email protected]> wrote: > Can I safely assume that my problem could not have been caused by a poorly > configured WFS service on my side and that the WFS 1.0.0 spec really > requires the PropertyName to be part of the BBOX filter?
Yes, the WFS 1.0 spec demands a PropertyName in the BBOX filter. On a side note, MapSever should never have worked with your default Protocol.WFS configuration, unless you had set gml_geometries in your mapfile metadata to "the_geom". Looks like MapServer requires the PropertyName, but doesn't respect the value you provide there. > In other words: > could you reproduce the error yourself using MapServer? Yes. See this GET request: http://demo.mapserver.org/cgi-bin/wfs?SERVICE=WFS&VERSION=1.0.0&REQUEST=getfeature&TYPENAME=continents&FILTER=%3CFilter%3E%3CBBOX%3E%3Cgml:PropertyName%3Ethe_geom%3C/gml:PropertyName%3E%3Cgml:Box%3E%3Cgml:coordinates%3E16.0,47.0%2016.1,47.1%3C/gml:coordinates%3E%3C/gml:Box%3E%3C/BBOX%3E%3C/Filter%3E The PropertyName is set to the_geom. But MapServer still returns Austria correctly, although the geometryName is msGeometry and not the_geom. But only looking at it from an XML schema validation perspective, MapServer is right to complain about an invalid filter in this request: http://demo.mapserver.org/cgi-bin/wfs?SERVICE=WFS&VERSION=1.0.0&REQUEST=GetFeature&TYPENAME=continents&FILTER=%3CFilter%3E%3CBBOX%3E%3Cgml:Box%3E%3Cgml:coordinates%3E16.0,47.0%2016.1,47.1%3C/gml:coordinates%3E%3C/gml:Box%3E%3C/BBOX%3E%3C/Filter%3E On the other hand, MapServer should not complain about this request: http://demo.mapserver.org/cgi-bin/wfs?SERVICE=WFS&VERSION=1.1.0&REQUEST=GetFeature&TYPENAME=continents&SRSNAME=CRS:84&FILTER=%3CFilter%3E%3CBBOX%3E%3Cgml:Box%3E%3Cgml:coordinates%3E16.0,47.0%2016.1,47.1%3C/gml:coordinates%3E%3C/gml:Box%3E%3C/BBOX%3E%3C/Filter%3E It is WMS 1.1, the PropertyName is not mandatory. Andreas. > On 18/07/11 05:46, Andreas Hocevar wrote: >> >> Hi Roald, >> >> please let me know http://trac.osgeo.org/openlayers/ticket/3415 looks >> good to you - with that patch applied, you should be able to use your >> old configuration again, unless you set the protocol version to 1.1.0 >> (default is 1.0.0), in which case the requirement for a propertyName >> in the BBOX filter would come from the MapServer bug >> http://trac.osgeo.org/mapserver/ticket/3955. >> >> Andreas. >> >> On Sun, Jul 17, 2011 at 12:41 AM, Roald de Wit<[email protected]> wrote: >>> >>> Hey Andreas, >>> >>> Thanks, it works again when adding a geometryName to the protocol. >>> >>> 2 things: >>> 1) this does change behaviour, so it would be important to document it >>> clearly in the release notes for 2.11 (if this makes it in there) >>> 2) in the comments in ticked #3368, bartvde says: >>> >>> "Maybe there is a WFS out there that needs PropertyName, but ignores >>> its value for WFS 1.0. But I think we can safely ignore until >>> someone files a bug with a case that fails. " >>> >>> >>> This is exactly the case for me with MS 6.0.x and 5.6.x too. MS needs a >>> PropertyName (can be anything so it seems!) and gives an error if it is >>> missing. >>> >>> It's a small change for me to add geometryName (or featureNS) to the >>> protocol options. I'm just slightly concerned that there are more people >>> out >>> there currently only using featurePrefix and featureType that might not >>> understand why their requests suddenly fail when using the new WFS >>> protocol >>> behaviour. >>> >>> Cheers, >>> >>> Roald >>> >>> On 17/07/11 03:24, Andreas Hocevar wrote: >>>> >>>> Hey Roald, >>>> >>>> what you describe is exactly the new expected behavior. However, if you >>>> add either a featureNS or a geometryName, things should be working >>>> again. >>>> >>>> Let me know if you still have concerns, or if and where you think this >>>> should be better documented. If the former, do you have a suggestion how >>>> we >>>> could better solve this? Because the intention of the change you suffer >>>> from >>>> was to make dynamic WFS configuration simpler. >>>> >>>> Andreas. >>>> >>>>> On Jul 16, 2011 2:38 PM, "Roald de Wit"<[email protected] >>>>> <mailto:[email protected]>> wrote: >>>>> >>>>> Hey Andreas, >>>>> >>>>> It's true that I don't set the FeatureNS. However, I *do* set the >>>>> featureType *and* the featurePrefix in my protocol options, which used >>>>> to be >>>>> sufficient. >>>>> If you take the wfs-protocol.html example from trunk and replace >>>>> featureNS with: featurePrefix: 'topp', you'll see that the PropertyName >>>>> disappears from the WFS request (in the BBOX filter). >>>>> >>>>> Is this expected behaviour, ie: can I not use featurePrefix anymore or >>>>> is >>>>> it a bug? >>>>> >>>>> Thanks, Roald >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> On 16/07/11 19:59, Andreas Hocevar wrote: >>>>> >>>>> > >>>>> > >>>>> > Hey Roald, >>>>> > >>>>> > the PropertyName will still always be set, unless you have a >>>>> poorly configured ... >>>>> >>>>> >> On Jul 16, 2011 8:25 AM, "Roald de Wit"<[email protected] >>>>> <mailto:[email protected]> <mailto:[email protected] >>>>> <mailto:[email protected]>>> wrote: >>>>> >> >>>>> >> H... >>>>> >>>>> [email protected]<mailto:[email protected]> >>>>> <mailto:[email protected]<mailto:[email protected]>> >>>>> >>>>> >>>>> >> http://lists.osgeo.org/mailman/listinfo/openlayers-dev >>>>> >>>>> >>>>> _______________________________________________ >>>>> Dev mailing list >>>>> [email protected]<mailto:[email protected]> >>>>> http://lists.o... >>>>> >>> _______________________________________________ >>> Dev mailing list >>> [email protected] >>> http://lists.osgeo.org/mailman/listinfo/openlayers-dev >>> >> >> > > _______________________________________________ > Dev mailing list > [email protected] > http://lists.osgeo.org/mailman/listinfo/openlayers-dev > -- Andreas Hocevar OpenGeo - http://opengeo.org/ Expert service straight from the developers. _______________________________________________ Dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/openlayers-dev
