Incorrect case in XML syntax for gml:null in a getfeature response

[nsavard]

The syntax for the XML tag should be gml:Null instead of gml:null based on ​http://schemas.opengis.net/gml/3.1.1/base/basicTypes.xsd

The request to output this response is:

​http://gw.mapgears.com:8500/cgi-bin/mswfs110_ogc_cite?&service=WFS&version=1.1.0&request=GetFeature&typename=sf:PrimitiveGeoFeature&bbox=0,0,0.01,0.01&namespace=xmlns

[tomkralidis]

Norm: fixed in r9827. Can you check and verify?

[nsavard]

Yes Tom, I'll check that.

[assefa]

ref: CITE test GetFeature-Get Test wfs:wfs-1.1.0-Basic-GetFeature-tc8.3

Note that the gml:null is valid for gml2.x (wfs1.0) (​http://schemas.opengis.net/gml/2.1.1/feature.xsd)

gml:Null is for gml3.x (wfs1.1) (​http://schemas.opengis.net/gml/3.1.1/base/basicTypes.xsd)

Another note the CITE test uses the GML SF-0 profile whne validating outputs (​http://schemas.opengis.net/gml/3.1.1/profiles/gmlsfProfile/1.0.0/gmlsf.xsd). In this profile gml:boundedBy can only have an Envelop elelement (no Null element). I believe we should generate an Envelop element with the current bbox of the map and srs instead of the Null element to conform to the GML level 0 profile

[nsavard]

That makes sense for me.

[nsavard]

Re #comment:3

Maybe we're waiting for each other to act on this issue. Tom, is it you who is going to fix this issue?

[tomkralidis]

Norm: thanks for the clarification.

Notes:

• in GML SF-0, gml:boundedBy is an optional element. We *could* ignore printing this out when using this profile in this case

Assefa: implementation questions:

• should we generate gml:Envelope with map.extent/srs instead of NULL for GML SF-0? Or not print out gml:boundedBy (for cases of no data returned)?
• how do we know whether the WFS client is asking for GML SF-0 as Norm states above, as opposed to plain old GML 3?
• here's the logic I'm proposing
  • if GML2 print out gml:null
  • if GML3 print out gml:Null
  • if GML SF-0 print out gml:boundedBy/gml:Envelope with map.extent/srs OR don't print gml:boundedBy at all

Comments?

[assefa]

I was under the impression that the current GML3 output of MapServer is based on GML SF-0. Isn't that the case? So users should always expect GML SF-0.

If so we should also send something valid in GML SF-0 for cases where no features is found.

[sdlime]

Yup, GML3 driver was based on the simple features profile (I thought). XML sucks...

Steve

[tomkralidis]

Update: it turns out GML SF-0 _does_ support gml:Null as well. So we don't have to worry about writing out gml:Envelope.

Here's the proposed patch, then:

Index: mapwfs.c
===================================================================
--- mapwfs.c    (revision 10152)
+++ mapwfs.c    (working copy)
@@ -2225,7 +2225,10 @@
     
     if (((iNumberOfFeatures==0) || (maxfeatures == 0)) && iResultTypeHits == 0) {
       msIO_printf("   <gml:boundedBy>\n"); 
-      msIO_printf("      <gml:Null>missing</gml:Null>\n");
+      if(outputformat == OWS_GML3)
+        msIO_printf("      <gml:Null>missing</gml:Null>\n");
+      else
+        msIO_printf("      <gml:null>missing</gml:null>\n");
       msIO_printf("   </gml:boundedBy>\n"); 
     }

Comments before I commit?

[sdlime]

Should this be implemented as a function in mapgml.c instead? Something like:

void gmlWriteNullFeature(FILE *stream, int outputformat) [
  msIO_fprintf(stream, "   <gml:boundedBy>\n");
  if(outputformat == OWS_GML3)
    msIO_fprintf(stream, "      <gml:Null>missing</gml:Null>\n");
  else
    msIO_fprintf(stream, "      <gml:null>missing</gml:null>\n");
  msIO_fprintf(stream, "   </gml:boundedBy>\n");
}

Just wondering if we'd use it elsewhere...

Steve

[assefa]

Tom,

I am looking into ​http://schemas.opengis.net/gml/3.1.1/base/basicTypes.xsd and from what I can see boundedBy can only have a gml:Envelope

Is that correct or is this the correct xsd?

[tomkralidis]

Replying to assefa:

Tom,

I am looking into ​http://schemas.opengis.net/gml/3.1.1/base/basicTypes.xsd and from what I can see boundedBy can only have a gml:Envelope

Is that correct or is this the correct xsd?

I don't see it in ​http://schemas.opengis.net/gml/3.1.1/base/basicTypes.xsd, but now I do see it in ​http://schemas.opengis.net/gml/3.1.1/profiles/gmlsfProfile/1.0.0/gmlsf.xsd (something with my XMLSpy cache).

Then again, I don't see any explicit reference to using the gmlsf schemas or namespaces in the codebase. Are we really using gmlsf or not?

[assefa]

Do you mean the xsd link that we are referring in the code (gml/3.1.1/base/gml.xsd)? I think that should be changed to the gml sf-0 but I have not checked if our output is compatible with the SF-0 schema.

[tomkralidis]

Replying to assefa:

Do you mean the xsd link that we are referring in the code (gml/3.1.1/base/gml.xsd)? I think that should be changed to the gml sf-0 but I have not checked if our output is compatible with the SF-0 schema.

I wonder how much change that would cause.

Nevertheless, looking again at ​http://schemas.opengis.net/gml/3.1.1/profiles/gmlsfProfile/1.0.0/gmlsf.xsd, you will see that gml:boundedBy only allows for gml:Envelope at the _feature_ level, as opposed to featurecollection level, per this ticket.

[assefa]

I did not realize this. I won't be able to get to the cite tests again before few weeks and at time I will see of It passes these particular tests. I have no more objection to your proposed commits. I will comment in this bug again if I have trouble with the tests (understanding if the cite tests or MapServer is at fault)

[tomkralidis]

Committed in r10158. This handles vanilla GML2/GML3. We don't _formally_ support gmlsf, so we can tackle that in a separate ticket.