[Geoserver-devel] wfs 1.1 cite test issues

GeoServer GeoServer Developer
1

A summary of where we stand on cite issues for wfs 1.1. Most of the failures found by victor were a result of the EntityGeneric layer and svn encoding issues. Basically when I checked it out on the server it somehow ended up with a non UTF encoding so comparisons with other UTF versions of the strings were failing. @Ben: I thought you had a way around this at some point?

Anyways, other then that there were two legit failures. Actually these were new cite tests recently added that require GML2 requests to return a mime type with the subtype included, rather than just text/xml. Easy fix.

Other then that there are still some issues that are "failing" but I don't consider blockers. They are explained below:

* wfs:wfs-1.1.0-Basic-GetFeature-tc42.4
---------------------------------------

fails on postgres versions 8.3 or greater. Been an open issue for quite some time.

* wfs:wfs-1.1.0-Basic-GetFeature-tc200.2
----------------------------------------

I was surprised buy this one. THe issue is that GEoServer complains about a validation failure with the following request:

...
<ogc:Filter xmlns:ogc="http://www.opengis.net/ogc&quot; xmlns:gml="http://www.opengis.net/gml&quot;&gt;
       <ogc:BBOX>
           <gml:Envelope srsName="urn:x-ogc:def:crs:EPSG:4326">
             <gml:lowerCorner>34 14</gml:lowerCorner>
             <gml:upperCorner>41 23</gml:upperCorner>
           </gml:Envelope>
       </ogc:BBOX>
     </ogc:Filter>
...

I could have sworn that the wfs and filter specs allowed one to leave off the PropertyName in a BBOX filter. But if i look at the filter schema [1] under BBOXType I indeed see that PropertyName is not optional.

So it would appear the test is in error here. Patching to include PropertyName it works fine.

* wfs:Transaction-idgen-use-existing
* wfs:Transaction-idgen-use-existing-update
----------------------------------------

These two tests require that the server accept feature ids from the client which we don't support. I have brought this up on the cite list but have gotten no feedback. But I argue strongly against a WFS supporting this even though there is a plan to implement this in GeoServer regardless.

If it is to be supported it has to be something that is settable on a feature type by feature type basis imo.

Also these tests were added sometime recently and not really cleared with any of the implementers afaik. If it was it was not made public.

So I don't think these should be considered blockers.

-Justin

[1] http://schemas.opengis.net/filter/1.1.0/filter.xsd

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

2

That sounds really great, thank you so much Justin :slight_smile:

I can see the update on the 2.0.2 tag so I am going to do one last build and
test and hopefully get 2.0.2 released today.

If anyone has any objection now is the time to say or hold your silence
forever :stuck_out_tongue:

Justin Deoliveira-6 wrote:

A summary of where we stand on cite issues for wfs 1.1. Most of the
failures found by victor were a result of the EntityGeneric layer and
svn encoding issues. Basically when I checked it out on the server it
somehow ended up with a non UTF encoding so comparisons with other UTF
versions of the strings were failing. @Ben: I thought you had a way
around this at some point?

Anyways, other then that there were two legit failures. Actually these
were new cite tests recently added that require GML2 requests to return
a mime type with the subtype included, rather than just text/xml. Easy
fix.

Other then that there are still some issues that are "failing" but I
don't consider blockers. They are explained below:

* wfs:wfs-1.1.0-Basic-GetFeature-tc42.4
---------------------------------------

fails on postgres versions 8.3 or greater. Been an open issue for quite
some time.

* wfs:wfs-1.1.0-Basic-GetFeature-tc200.2
----------------------------------------

I was surprised buy this one. THe issue is that GEoServer complains
about a validation failure with the following request:

...
<ogc:Filter xmlns:ogc="http://www.opengis.net/ogc&quot;
xmlns:gml="http://www.opengis.net/gml&quot;&gt;
       <ogc:BBOX>
           <gml:Envelope srsName="urn:x-ogc:def:crs:EPSG:4326">
             <gml:lowerCorner>34 14</gml:lowerCorner>
             <gml:upperCorner>41 23</gml:upperCorner>
           </gml:Envelope>
       </ogc:BBOX>
     </ogc:Filter>
...

I could have sworn that the wfs and filter specs allowed one to leave
off the PropertyName in a BBOX filter. But if i look at the filter
schema [1] under BBOXType I indeed see that PropertyName is not optional.

So it would appear the test is in error here. Patching to include
PropertyName it works fine.

* wfs:Transaction-idgen-use-existing
* wfs:Transaction-idgen-use-existing-update
----------------------------------------

These two tests require that the server accept feature ids from the
client which we don't support. I have brought this up on the cite list
but have gotten no feedback. But I argue strongly against a WFS
supporting this even though there is a plan to implement this in
GeoServer regardless.

If it is to be supported it has to be something that is settable on a
feature type by feature type basis imo.

Also these tests were added sometime recently and not really cleared
with any of the implementers afaik. If it was it was not made public.

So I don't think these should be considered blockers.

-Justin

[1] http://schemas.opengis.net/filter/1.1.0/filter.xsd

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------

_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

--
View this message in context: http://old.nabble.com/wfs-1.1-cite-test-issues-tp28615908p28616455.html
Sent from the GeoServer - Dev mailing list archive at Nabble.com.

3

Well actually it seems there are some minor issues with wcs 1.0.0 cite tests as well :frowning:

I think i am done working through the issues. I will send email and commit soon.

-Justin

On 10-05-19 8:26 PM, VT@anonymised.com wrote:

That sounds really great, thank you so much Justin :slight_smile:

I can see the update on the 2.0.2 tag so I am going to do one last build and
test and hopefully get 2.0.2 released today.

If anyone has any objection now is the time to say or hold your silence
forever :stuck_out_tongue:

Justin Deoliveira-6 wrote:

A summary of where we stand on cite issues for wfs 1.1. Most of the
failures found by victor were a result of the EntityGeneric layer and
svn encoding issues. Basically when I checked it out on the server it
somehow ended up with a non UTF encoding so comparisons with other UTF
versions of the strings were failing. @Ben: I thought you had a way
around this at some point?

Anyways, other then that there were two legit failures. Actually these
were new cite tests recently added that require GML2 requests to return
a mime type with the subtype included, rather than just text/xml. Easy
fix.

Other then that there are still some issues that are "failing" but I
don't consider blockers. They are explained below:

* wfs:wfs-1.1.0-Basic-GetFeature-tc42.4
---------------------------------------

fails on postgres versions 8.3 or greater. Been an open issue for quite
some time.

* wfs:wfs-1.1.0-Basic-GetFeature-tc200.2
----------------------------------------

I was surprised buy this one. THe issue is that GEoServer complains
about a validation failure with the following request:

...
<ogc:Filter xmlns:ogc="http://www.opengis.net/ogc&quot;
xmlns:gml="http://www.opengis.net/gml&quot;&gt;
        <ogc:BBOX>
            <gml:Envelope srsName="urn:x-ogc:def:crs:EPSG:4326">
              <gml:lowerCorner>34 14</gml:lowerCorner>
              <gml:upperCorner>41 23</gml:upperCorner>
            </gml:Envelope>
        </ogc:BBOX>
      </ogc:Filter>
...

I could have sworn that the wfs and filter specs allowed one to leave
off the PropertyName in a BBOX filter. But if i look at the filter
schema [1] under BBOXType I indeed see that PropertyName is not optional.

So it would appear the test is in error here. Patching to include
PropertyName it works fine.

* wfs:Transaction-idgen-use-existing
* wfs:Transaction-idgen-use-existing-update
----------------------------------------

These two tests require that the server accept feature ids from the
client which we don't support. I have brought this up on the cite list
but have gotten no feedback. But I argue strongly against a WFS
supporting this even though there is a plan to implement this in
GeoServer regardless.

If it is to be supported it has to be something that is settable on a
feature type by feature type basis imo.

Also these tests were added sometime recently and not really cleared
with any of the implementers afaik. If it was it was not made public.

So I don't think these should be considered blockers.

-Justin

[1] http://schemas.opengis.net/filter/1.1.0/filter.xsd

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------

_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

4

My wcs1.0 passed when I was testing it. But I am doing a full rebuild now and
I will run all test cases again :slight_smile:

Justin Deoliveira-6 wrote:

Well actually it seems there are some minor issues with wcs 1.0.0 cite
tests as well :frowning:

I think i am done working through the issues. I will send email and
commit soon.

-Justin

On 10-05-19 8:26 PM, VT@anonymised.com wrote:

That sounds really great, thank you so much Justin :slight_smile:

I can see the update on the 2.0.2 tag so I am going to do one last build
and
test and hopefully get 2.0.2 released today.

If anyone has any objection now is the time to say or hold your silence
forever :stuck_out_tongue:

Justin Deoliveira-6 wrote:

A summary of where we stand on cite issues for wfs 1.1. Most of the
failures found by victor were a result of the EntityGeneric layer and
svn encoding issues. Basically when I checked it out on the server it
somehow ended up with a non UTF encoding so comparisons with other UTF
versions of the strings were failing. @Ben: I thought you had a way
around this at some point?

Anyways, other then that there were two legit failures. Actually these
were new cite tests recently added that require GML2 requests to return
a mime type with the subtype included, rather than just text/xml. Easy
fix.

Other then that there are still some issues that are "failing" but I
don't consider blockers. They are explained below:

* wfs:wfs-1.1.0-Basic-GetFeature-tc42.4
---------------------------------------

fails on postgres versions 8.3 or greater. Been an open issue for quite
some time.

* wfs:wfs-1.1.0-Basic-GetFeature-tc200.2
----------------------------------------

I was surprised buy this one. THe issue is that GEoServer complains
about a validation failure with the following request:

...
<ogc:Filter xmlns:ogc="http://www.opengis.net/ogc&quot;
xmlns:gml="http://www.opengis.net/gml&quot;&gt;
        <ogc:BBOX>
            <gml:Envelope srsName="urn:x-ogc:def:crs:EPSG:4326">
              <gml:lowerCorner>34 14</gml:lowerCorner>
              <gml:upperCorner>41 23</gml:upperCorner>
            </gml:Envelope>
        </ogc:BBOX>
      </ogc:Filter>
...

I could have sworn that the wfs and filter specs allowed one to leave
off the PropertyName in a BBOX filter. But if i look at the filter
schema [1] under BBOXType I indeed see that PropertyName is not
optional.

So it would appear the test is in error here. Patching to include
PropertyName it works fine.

* wfs:Transaction-idgen-use-existing
* wfs:Transaction-idgen-use-existing-update
----------------------------------------

These two tests require that the server accept feature ids from the
client which we don't support. I have brought this up on the cite list
but have gotten no feedback. But I argue strongly against a WFS
supporting this even though there is a plan to implement this in
GeoServer regardless.

If it is to be supported it has to be something that is settable on a
feature type by feature type basis imo.

Also these tests were added sometime recently and not really cleared
with any of the implementers afaik. If it was it was not made public.

So I don't think these should be considered blockers.

-Justin

[1] http://schemas.opengis.net/filter/1.1.0/filter.xsd

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------

_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------

_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

--
View this message in context: http://old.nabble.com/wfs-1.1-cite-test-issues-tp28615908p28616643.html
Sent from the GeoServer - Dev mailing list archive at Nabble.com.

5

We may have a version conflict with the tests... are you running hte trunk version of the wcs tests from svn? Or an older one that was posted?

On 10-05-19 9:01 PM, VT@anonymised.com wrote:

My wcs1.0 passed when I was testing it. But I am doing a full rebuild now and
I will run all test cases again :slight_smile:

Justin Deoliveira-6 wrote:

Well actually it seems there are some minor issues with wcs 1.0.0 cite
tests as well :frowning:

I think i am done working through the issues. I will send email and
commit soon.

-Justin

On 10-05-19 8:26 PM, VT@anonymised.com wrote:

That sounds really great, thank you so much Justin :slight_smile:

I can see the update on the 2.0.2 tag so I am going to do one last build
and
test and hopefully get 2.0.2 released today.

If anyone has any objection now is the time to say or hold your silence
forever :stuck_out_tongue:

Justin Deoliveira-6 wrote:

A summary of where we stand on cite issues for wfs 1.1. Most of the
failures found by victor were a result of the EntityGeneric layer and
svn encoding issues. Basically when I checked it out on the server it
somehow ended up with a non UTF encoding so comparisons with other UTF
versions of the strings were failing. @Ben: I thought you had a way
around this at some point?

Anyways, other then that there were two legit failures. Actually these
were new cite tests recently added that require GML2 requests to return
a mime type with the subtype included, rather than just text/xml. Easy
fix.

Other then that there are still some issues that are "failing" but I
don't consider blockers. They are explained below:

* wfs:wfs-1.1.0-Basic-GetFeature-tc42.4
---------------------------------------

fails on postgres versions 8.3 or greater. Been an open issue for quite
some time.

* wfs:wfs-1.1.0-Basic-GetFeature-tc200.2
----------------------------------------

I was surprised buy this one. THe issue is that GEoServer complains
about a validation failure with the following request:

...
<ogc:Filter xmlns:ogc="http://www.opengis.net/ogc&quot;
xmlns:gml="http://www.opengis.net/gml&quot;&gt;
         <ogc:BBOX>
             <gml:Envelope srsName="urn:x-ogc:def:crs:EPSG:4326">
               <gml:lowerCorner>34 14</gml:lowerCorner>
               <gml:upperCorner>41 23</gml:upperCorner>
             </gml:Envelope>
         </ogc:BBOX>
       </ogc:Filter>
...

I could have sworn that the wfs and filter specs allowed one to leave
off the PropertyName in a BBOX filter. But if i look at the filter
schema [1] under BBOXType I indeed see that PropertyName is not
optional.

So it would appear the test is in error here. Patching to include
PropertyName it works fine.

* wfs:Transaction-idgen-use-existing
* wfs:Transaction-idgen-use-existing-update
----------------------------------------

These two tests require that the server accept feature ids from the
client which we don't support. I have brought this up on the cite list
but have gotten no feedback. But I argue strongly against a WFS
supporting this even though there is a plan to implement this in
GeoServer regardless.

If it is to be supported it has to be something that is settable on a
feature type by feature type basis imo.

Also these tests were added sometime recently and not really cleared
with any of the implementers afaik. If it was it was not made public.

So I don't think these should be considered blockers.

-Justin

[1] http://schemas.opengis.net/filter/1.1.0/filter.xsd

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------

_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------

_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.