RE: http:location grammar

Should be 

	elementName ::= '{' NCName '}'

We simply ignore the namespace and just key off the local name.  IIRC, we
figured the added complexity isn't worth the likelihood that someone will
have data that includes both a my:name and a your:name yet stills fall
within the constraints of the IRI style.

Perhaps we need to clarify that "local name" does indeed imply "ignore the
namespace" in [1].

[1]
http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl20/wsdl20-adjuncts.html
?content-type=text/html;%20charset=utf-8#_http_operation_location_cited_ser

Jonathan Marsh - http://www.wso2.com - http://auburnmarshes.spaces.live.com
 

> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On
> Behalf Of Philippe Le Hegaret
> Sent: Friday, January 12, 2007 7:24 AM
> To: www-ws-desc
> Subject: http:location grammar
> 
> 
> Using EBNF, the grammar to be applied to the Infoset value of
> http:location is:
> 
> httpLocation ::= CharData? (( openBrace | closeBrace | elementName )
> CharData?)*
> 
> CharData ::= [^{}]*
> 
> openBrace ::= '{{'
> 
> closeBrace ::= '}}'
> 
> elementName ::= '{' QName '}'
> 
> The lexical space of QName is defined in the namespace specification and
> is also equivalent to xsd:QName. I believe we only support element
> declarations with namespaces so this explains the use QName instead of
> Name.
> 
> This gives the following results (same as Jonathan's, with an additional
> one):
> 
> "{town}"       > {town}           >  "Paris"
> "{{town}}"     > {{,town,}}       >  "{town}"
> "{{{town}}}"   > {{,{town},}}     >  "{Paris}"
> "{{{{town}}}}" > {{,{{,town,}},}} >  "{{town}}"
> "{{town}"      > {{,town,}        >  ERROR
> "{{{town}"     > {{,{town}        >  "{Paris"
> "{town}}"      > {town},}         >  ERROR
> "{town}}}"     > {town},}}        >  "Paris}"
> "{t{o}wn}"     > {t{o}wn}         >  ERROR
> 
> I believe this is an LR(k) grammar by the way, with an unlimited k
> (QName is unlimited). It's possible k could be bounded to 1 by adding
> more states but we don't need to worry about that here.
> 
> Now, unless I missed a sentence or two in the specifications, here are
> two questions:
> 
> With:
> "{town}"       > {town}           >  "Paris"
> "{tr:town}"       > {tr:town}           >  "Paris"
> 
> What are the namespaces associated with "town" or "tr:town"? We store
> {http location} in the content model as an xs:anyURI but we don't keep
> the namespaces in scope around despite the fact that further processing
> might need them to resolve the qnames. Should we clarify that namespace
> resolution is done according to the namespaces in scope and/or add a
> property to keep those namespaces in scope around? By just using the
> 
> MessageTest-1G defines the following:
> 
> <xs:schema elementFormDefault="qualified"
>       xmlns:xsmt="http://example.org/message-test/xsd">
>  <xs:element name="EchoInt" type="xsmt:intStruct"/>
>  <xs:complexType name="intStruct">
>      <xs:sequence>
>          <xs:element name="int" type="xs:int"/>
>      </xs:sequence>
>  </xs:complexType>
> </xs:schema>
> [...]
> <interface>
>  <operation name="EchoInt"
>       style="http://www.w3.org/2006/01/wsdl/style/iri">
>       <input element="xsmt:EchoInt"/>
>       <output element="xsmt:echoIntReturn"/>
>  </operation>
> </interface>
> [...]
> <binding>
>     <operation ref="tns:EchoInt"
>       wsoap:action="http://example.org/message-test/action/EchoInt"
>       whttp:location="?op=EchoInt;int={int};" />
> </binding>
> 
> Shouldn't it be {xsmt:int} instead?
> 
> 
> Philippe
> 
>