[Requests] CSW2 AP ISO (RFPC 35) Response

Fri May 19 12:50:16 EDT 2006

PART A
------
 1. Evaluator: Richard Martell <rmartell AT galdosinc DOT com>
               Galdos Systems, Inc.

 2. Submission: 
    OGC Request 35: OpenGIS Catalogue Services Specification 2.0.1 (with 
    Corrigendum) - ISO Metadata Application Profile (CSW2 AP ISO): Request 
    for Public Comments  

================================================================================
PART B
------

1.1 Specification Section number: 6.2.3 - Harvest metadata
1.2 Criticality: MINOR
1.3 Comments/justifications for changes:

    Consider relaxing the pre-condition that the XML resource be valid--the 
    service will attempt to validate the request anyway, no?
    
    Amend the post-condition as follows: 
    "If the request is processed successfully, the XML resource is inserted..."


2.1 Specification Section number: 7.2 - Catalogue information model
2.2 Criticality: EDITORIAL
2.3 Comments/justifications for changes:

    The introductory clause should mention ISO 19139, the XML realization of 
    ISO 19115 that is used by this profile.


3.1 Specification Section number: 7.2.1.2 - Service
3.2 Criticality: MINOR
3.3 Comments/justifications for changes:

    In Table 3, couplingType seems unnecessary: either the service description 
    refers to data set descriptions (tightly-coupled case) or it doesn't. 
    Alternatively, replace it with a simple boolean type, where 
    tightlyCoupled="false" means the service can operate on other congruent 
    data, not just the one(s) referenced.


4.1 Specification Section number: 7.2.3.1 - OGC core queryable properties
4.2 Criticality: MAJOR
4.3 Comments/justifications for changes:

    In the CSW part (OGC 04-021r3, clause 10) there is no real distinction 
    between common queryable and returnable elements: the common facets 
    declared for the csw:Record type (04-021r3, 10.2.5) may be both queried 
    _and_ returned (these are mapped to supported record representations).
    
    In Table 10, also include mappings to ISO 19139 elements (i.e., normative 
    XML representation). Or just replace the mappings to 19115, since clients 
    don't query abstract information models but concrete realizations thereof 
    (ISO 19139). Or does this mean the 19139 representations themselves cannot 
    be queried? Can query filters include XPath expressions that reflect the 
    19139 infoset? Please clarify.


5.1 Specification Section number: 7.2.4 - Additional search properties
5.2 Criticality: EDITORIAL
5.3 Comments/justifications for changes:

    The filter example on p. 44 is incorrect: csw:service not defined in the 
    "http://www.opengis.net/cat/csw" namespace (but the "csw" prefix is not 
    bound, so the exact target namespace is unknown in this example).


6.1 Specification Section number: 7.4 - Result sets
6.2 Criticality: MAJOR
6.3 Comments/justifications for changes:

    With the latest OGC Catalogue revision (2.0.1), it is no longer necessary 
    to substitute for csw:AbstractRecord. The ISO 19139 schemas may be used 
    directly without a "wrapper" schema. 
    
    The csw:IsoRecord element is not declared anywhere. In Table 16, replace 
    <csw:IsoRecord> with <gmd:MD_Metadata> or some allowable substitution 
    (where gmd is bound to the "http://www.isotc211.org/2005/gmd" namespace). 

    Also, perhaps allow the use of <gmx:MX_DataSet> to represent a dataset 
    description accessible from some external repository managed by another 
    party.


7.1 Specification Section number: 8.2 - Interface specifications
7.2 Criticality: EDITORIAL
7.3 Comments/justifications for changes:

    This clause should simply specify any extensions or restrictions to the base 
    specifications, not duplicate material from those documents (less possibility 
    of inconsistencies or errors). Profiles are not stand-alone documents.


8.1 Specification Section number: Annex C - WSDL specifications
8.2 Criticality: MINOR
8.3 Comments/justifications for changes:

    WSDL 2.0 is now a W3C CR (WSDL 1.1 was a Microsoft/IBM spec). Consider also 
    including WSDL 2.0 definitions in preference to legacy definitions.
    See <http://www.w3.org/TR/wsdl20/>.


9.1 Specification Section number: Annex D - XML Schemas (non-standard views)
9.2 Criticality: MAJOR
9.3 Comments/justifications for changes:

    The brief and summary view elements (<rsb:MD_Metadata> and <rsb:MD_Metadata>, 
    respectively) do not include _any_ elements from the 19139 namespace, nor 
    can they substitute for <gmd:MD_Metadata>. Instead, new elements are 
    declared in the profile namespace. This quirk is likely due to the fact that 
    the component elements of the gmd:MD_Metadata_Type definition are declared 
    locally, not globally--a shortcoming of the 19139 schemas that inhibits 
    profiling.

    Rather than defining new content models (non-standard views), consider just 
    aligning the brief and summary views with the minimal and core subsets 
    (19139, D.1 and D.2). Yes, XML Schema validation for these views then becomes 
    impractical, but other complementary validation mechanisms can be used 
    (e.g. Schematron, as in 19136).

    One could request that TC 211 either (1) change the content model of 
    gmd:MD_Metadata_Type to use global elements (so it can then be restricted), 
    or (2) add two new type definitions in the gmd namespace corresponding to 
    these subsets: gmd:MD_Metadata_Minimal_Type and gmd:MD_Metadata_Core_Type 
    as ready-made restrictions. Or both :-)
    
    I have submitted requests for these changes through the ISO 19139 issue 
    tracker (#23, #24): <http://eden.ign.fr/xsd/isotc211/19139>.


10.1 Specification Section number: Annex D - XML Schemas (services)
10.2 Criticality: MAJOR
10.3 Comments/justifications for changes:

    OGC 05-008 (OGC service capabilities) and the W3C WSDL specs seem adequate 
    for describing the computational and non-computational aspects of web services.
    ISO 19119 seems superfluous--it doesn't appear to express anything not 
    already captured by these existing specs. Please clarify how services.xsd 
    in Annex D relates to the harmonized OGC capabilities schemas. Is it an 
    alternative representation? A replacement? Is there a mapping between the 
    two?
  
    The latest ISO 19139 spec (final DTS, 2005-10) makes no reference whatsoever 
    to ISO 19119, so service descriptions in the 19139 namespaces seem rather 
    out of place. Also, setting the target namespace for services.xsd to a 19139 
    namespace violates the intent of conformance clauses A.3(4,5) in 19139:

    "New elements may not be added directly to the XML Schemas defined by this 
    Technical Specification."

    "Any new metadata entities should be added in their own namespace..."
  
    The service description elements should be assigned to another namespace  
    (e.g., "http://www.isotc211.org/2006/srv" if defined under the auspices of 
    TC 211)
  
    [Attachment: services2.xsd]
-------------- next part --------------
A non-text attachment was scrubbed...
Name: services2.xsd
Type: application/octet-stream
Size: 7243 bytes
Desc: services2.xsd
Url : https://mail.opengeospatial.org/mailman/private/requests/attachments/20060519/3aa05fd3/services2.obj