Re: [ws-rx] Re: [ws-rx-editors] possible bug in draft spec 'wsrm-1.1-spec-wd-05'(fwd)

[Anish Karmarkar <Anish.Karmarkar@oracle.com>]

Robin Cover wrote:
> Anish said:
> 
> 
>>Comments?
> 
> 
> I have no stake in how the TC's decision falls out with respect to
> one of these two (namespace URIs):
> 
> http://docs.oasis-open.org/wsrm/200510/
> http://docs.oasis-open.org/wsrm/200510
> 
> However, I am interested in current expert opinion about this design
> point, in the abstract.
> 
> Summary: per text below, there's an apparent bug in the document 
> published as
> 'http://www.oasis-open.org/apps/org/workgroup/ws-rx/download.php/14785/wsrm-1.1-spec-wd-05.pdf'
> because as the text stands, concatenation of the Namespace URI
> with "/" and "element name" yields malformed things like:
> 
> http://docs.oasis-open.org/wsrm/200510//SequenceAcknowledgement
> 
> Doug suggested removal of the trailing slash "/" from the
> namespace, while Anish suggested the alternative, viz.,
> changing the language about what's concatenated with the
> namespace URI.
> 
> One argument for retaining the namespace with "/" would be
> that this makes it easy to construct URIs from QNames, viz.,
> concatenate the namespace name with the localname.
> 

I'm no expert in this, but I know of two ways that people do QNames->URI 
mapping:
1) {namespace URI}localname invented by James Clark [1]
2) concatenation of NS URI with local name (as you mention above) used 
by RDF [2].

This was also a W3C TAG issue, not sure how it got resolved. But I 
*think* it got resolved by the TAG saying that a spec that uses QName 
refs should define the QName->URI mapping. AFAIK, the TAG did not 
specify any mapping.

 From a QName->URI perspective, if there isn't a trailing '/' in the NS, 
concatenating the NS name with localname still yields a valid URI. For 
example, "http://docs.oasis-open.org/wsrm/200510SequenceAcknowledgement"; 
is a valid URI. The problem occurs in the URI->QName round-tripping.

My .02

-Anish
--

[1] http://www.jclark.com/xml/xmlns.htm
[2] http://www.w3.org/TR/rdf-syntax-grammar/

> Comments?  Considerations in articulating a preference?
> 
> Trying to learn from the experts,
> Robin Cover
> 
> ---------- Forwarded message ----------
> Date: Fri, 07 Oct 2005 14:55:07 -0700
> From: Anish Karmarkar <Anish.Karmarkar@oracle.com>
> To: Doug Davis <dug@us.ibm.com>
> Cc: Robin Cover <robin@oasis-open.org>,
>      WS-RX Editors <ws-rx-editors@lists.oasis-open.org>
> Subject: Re: [ws-rx-editors] possible bug in draft spec
>     'wsrm-1.1-spec-wd-05'
> 
> Personally, I don't care how this is solved as long as the bug that 
> Robin points out is fixed.
> 
> But, given that the NS was explicitly approved by the TC (and I'm not 
> sure if the TAB AIR document requires the NS to have a trailing '/'), I 
> think removing the "concatenated with the '/' character ..." part would 
> be the easier way forward.
> 
> I.e. change:
> "If an action URI is used, and one is not already defined per the rules 
> of the WSAddressing specification [WS-Addressing], then the action URI 
> MUST consist of the reliable messaging namespace URI concatenated with 
> the "/" character and the element name."
> 
> to:
> "If an action URI is used, and one is not already defined per the rules 
> of the WSAddressing specification [WS-Addressing], then the action URI 
> MUST consists of the reliable messaging namespace URI concatenated with 
> the element name."
> 
> BTW, the sentences is ambiguous about which element it is talking about. 
>   But that is a separate issue. It is also unclear as to what happens 
> when there is more than one WSRM header in the same message (eg: SeqAck 
> and AckRequested for two separate Sequences).
> 
> Comments?
> 
> -Anish
> --
> 
> Doug Davis wrote:
> 
>>Editors/Robin, Any objections to removing the / from the namespace?
>>-Doug
>>
>>*Robin Cover <robin@oasis-open.org>*
>>
>>10/07/2005 01:11 PM
> 
> 
>>To WS-RX Editors <ws-rx-editors@lists.oasis-open.org>
>>cc Robin Cover <robin.cover@oasis-open.org>
>>Subject [ws-rx-editors] possible bug in draft spec 'wsrm-1.1-spec-wd-05'
>>
>>http://www.oasis-open.org/apps/org/workgroup/ws-rx/download.php/14785/wsrm-1.1-spec-wd-05.pdf
>>
>>says:
>>
>>====================================================================
>>
>>1.3 Namespace
>>The XML namespace [XML-ns] URI that MUST be used by
>>implementations of this specification is:
>>http://docs.oasis-open.org/wsrm/200510/
>>[...]
>>
>>If an action URI is used, and one is not already defined per the rules
>>of the WS-Addressing specification [WS-Addressing], then the action  
>>URI MUST consist of the reliable messaging namespace URI concatenated
>>with the "/" character and the element name. For example:
>>
>>http://docs.oasis-open.org/wsrm/200510/SequenceAcknowledgement
>>
>>====================================================================
>>
>>If the "XML namespace URI" is (character-for-character)
>>
>>http://docs.oasis-open.org/wsrm/200510/
>>
>>then the "namespace URI concatenated with the '/' character and
>>the element name" would yield:
>>
>>http://docs.oasis-open.org/wsrm/200510/ + '/' + element name
>>
>>thus, e.g., (not what we want, I assume):
>>
>>http://docs.oasis-open.org/wsrm/200510//SequenceAcknowledgement
>>
>>Presumably: change the Namespace URI or nuke "concatenated with..."
>>
>>
>>- Robin Cover
> 
>