> >What rationale does BP give for saying that putting namespace
> >declarations on soap:Body is a bad thing? Seems silly to me.
>
> See here:
>
> http://www.ws-i.org/Profiles/Basic/2003-08/BasicProfile-1.0a.h
> tml#refinement34313448
>
> R2716 A document-literal binding in a DESCRIPTION MUST NOT have the
> namespace attribute specified on contained soapbind:body,
> soapbind:header, soapbind:headerfault and soapbind:fault elements.
>
> ...
>
> In a document-literal SOAP binding, the serialized element child of
> the soap:Body gets its namespace from the targetNamespace of the
> schema that defines the element. Use of the namespace attribute of
> the soapbind:body element would override the element's namespace.
> This is not allowed by the Profile.
Good point. Thanks for leading Axis one step forward to WS-I BP 1.0. :-)
>
> Note that R2717 says the opposite for rpc/literal.
>
> BTW, I don't have an easy way to track the code changes. Are
> y'all creating tests for the bugs you close? That way the bug
> will never be reintroduced.
>
Testing Java->WSDL (and even WSDL->Java) seems quite tough in terms of
artifacts, not runtime workability. Probably we need a testing framework for
a WSDL processor, particularly, walking with WS-I Profiles. How about
inventing WSUnit (WebServicesUnit)? (Or already exists?) For example,
Having Java classes
->
Processing them to a WSDL
->
Checking certain assertions to the generated WSDL
and vice versa,
Having a WSDL
->
Processing it to Java classes
->
Checking certain assertions to the generated classes
Your comments and opinions are welcome all the time.
Regards,
Ias
|