[CITE-Forum] WFS 1.1.0 CITE tests: multiplegeometrytypes ona single layer

Richard Martell rmartell at galdosinc.com
Fri Apr 25 16:17:47 EDT 2008


Hi all,


These are the specs used to assess WFS 1.1 conformance; that is, WFS-1.1 
PLUS all of its normative dependencies:

- WFS 1.1 (OGC 04-094)
- OWS Common 1.0 (OGC 05-008) *deprecated*
- Filter 1.1 (OGC 04-095)
- GML 3.1 (OGC 03-105r1) *deprecated*


Yes, there's GML 3.1 lurking at the bottom of the stack. The problem 
is that neither WFS-1.1 nor Filter-1.1 impose any constraints on 
GML usage; this implies _full_ support for GML. 

Now, no one in the known universe comes close to grokking the complete
GML element set (nor cares to, I'd wager). So for the purposes of 
conformance testing the GMLSF profile (OGC 06-049r1) was adopted in 
order to cut GML down to size; the test data were structured accordingly.
As pointed out previously, GMLSF explicitly allows multiple geometry 
properties (but only one occurrence of each at level SF-0); that is, 
the service is expected to handle such features. 

The profile could, of course, be revised to impose additional 
restrictions (at level SF-0?)


--
Richard


> -----Original Message-----
> From: 
> cite-forum-bounces+rmartell=galdosinc.com at lists.opengeospatial
> .org 
> [mailto:cite-forum-bounces+rmartell=galdosinc.com at lists.openge
> ospatial.org] On Behalf Of Ian Painter
> Sent: Friday, 25 April, 2008 10:05
> To: Raj Singh; David Arctur
> Cc: Kralidis, Tom [Burlington]; 
> cite-forum at lists.opengeospatial.org; Yewondwossen Assefa
> Subject: Re: [CITE-Forum] WFS 1.1.0 CITE tests: 
> multiplegeometrytypes ona single layer
> 
> All,
> 
> Having just got our WFS through the 1.1.0 compliance I can totally
> relate to the amount of effort / pain involved in getting 1.1.0
> compliance. The WFS 1.1.0 tests are very comprehensive and adhere very
> tightly to the spec (which is good thing). Given this, the development
> required to get 1.1.0 compliance is considerable and shouldn't be
> underestimated. It took us nearly 4 man months of development 
> to get our
> WFS through despite our WFS already being 1.0 compliant. 
> 
> Back to my concern ... in Europe there are numerous widely adopted
> schemas that have more than one geometry per feature and adhere to
> simple features. If we were to say that a WFS is 1.1.0 compliant then
> people will naturally assume that it fully supports simple 
> features and
> therefore supports multiple geometry within a single feature. My
> interpretation of the spec is that support for documents with only 1
> geometry per feature isn't support for simple features. I'm afraid I
> agree with Richard that either the Level 0 constraints need to further
> restricted or WFS providers need to take the development hit 
> (as we did)
> in order to pass the tests. 
> 
> I appreciate that this is quite a hard-line response but we must never
> get into situation of 'my WFS doesn't support X therefore can 
> you take X
> out of the compliance test'.
> 
> Ian
> 
> -----Original Message-----
> From:
> cite-forum-bounces+ian.painter=snowflakesoftware.com at lists.ope
> ngeospatia
> l.org
> [mailto:cite-forum-bounces+ian.painter=snowflakesoftware.com at l
> ists.openg
> eospatial.org] On Behalf Of Raj Singh
> Sent: 24 April 2008 22:04
> To: David Arctur
> Cc: 'Kralidis, Tom [Burlington]'; cite-forum at lists.opengeospatial.org;
> 'Yewondwossen Assefa'
> Subject: Re: [CITE-Forum] WFS 1.1.0 CITE tests: multiple geometrytypes
> ona single layer
> 
> This statement:
> > "(c) features may have any number of geometric properties" (p. 20)
> 
> says to me that a compliant document could have 0 or more geometric
> properties. There is no statement regarding the minimum 
> number or types
> of geometries required to be supported. Therefore a document with a
> single geometric property should pass all required tests (a document
> with 0 geometric properties should also pass, but I think 
> that is "a bad
> thing"). In this case I don't think the test data set matches what the
> specification states.
> 
> ---
> Raj



More information about the CITE-Forum mailing list