Re: [regrep-comment] regrep-ws-profile: Clarafication required aboutid attribute rules

[Farrukh Najmi <farrukh@wellfleetsoftware.com>]

andreas.veithen@gfi.be wrote:
> Dear all,
>
> The "ebXML Registry profile for Web Services" specification (as of version 
> 1.0 Draft 3) contains rules for the id attribute values of different 
> registry objects mapped from a WSDL. For example, section 4.1.1 specifies 
> that
>
> "The id attribute value of the rim:Service MUST have as prefix the 
> targetNamespace for the wsdl:service element, followed by a suffix of 
> “:service:<service name>” where <service name> MUST be the value of the 
> name attribute of the wsdl:service element."
>
> I think that this statement is inconsistent and/or needs clarification. 
> The reason is that the id in the registry must be a URN, whereas the 
> targetNamespace is a URI. Since the concept of URI is a superset of the 
> concept of URN, it is inconsistent to require constructing a URN by 
> prefixing it with a given URI, unless a prescription is given to map URIs 
> to URNs. For example, "http://www.acme.org/service"; is a valid value for 
> targetNamespace. Naively applying the above rule would give something like 
> "http://www.acme.org/service:service:MyService";, which of course is not a 
> valid URN. Note that in this case, the OMAR implementation produces an id 
> such as "urn:http:__www_acme_org_service:service:MyService", i.e. OMAR 
> actually has some rules to map from URIs to URNs (which by the way don't 
> guarantee uniqueness). I think that the regrep-ws-profile specs should 
> clearly specify the mapping to be used or alternatively make reference to 
> an existing specification defining this type of mapping.
>   

Hi Andreas,

You are correct that the current spec is under-specified in this regard 
and does not define the rules for generating a valid URN from the 
potentially invalid URN generated by the spec-defined algorithm. As you 
observe from the freebXML Registry 3.0 implementation this is a known 
issue that needs to be addressed with some additional spec content. We 
will fix this issue in the next version of the spec.

> PS: Could you also consider adding to regrep-rim a normative reference 
> defining the concept of URN (RFC2141?)?
>   

That is a good suggestion and will be addressed in the next errata and 
version of ebRIM.

Thanks again for your valuable comments.

-- 
Regards,
Farrukh

Web: http://www.wellfleetsoftware.com