- From: Jean-Jacques Moreau <moreau@crf.canon.fr>
- Date: Tue, 02 Jul 2002 16:37:44 +0200
- To: "Liu Kevin" <kevin.liu@sap.com>
- CC: "'Sanjiva Weerawarana'" <sanjiva@watson.ibm.com>, "WS-Desc WG (Public)" <www-ws-desc@w3.org>
Kevin, thanks for pointing these out. I've have left the most controversial ones, done the others. Jean-Jacques. "Liu, Kevin" wrote: > MINOR/EDITORIAL ERRORS > >>>Section 2 (text above the box for "issue-require-targetnamspace") > indicates that targetNamespace is required, whereas section 1.2 (if we > decide to keep it) the grammar still shows that defintions@targetNamespace > is optional --- the ? after targetNamespace should be removed Done. Well spotted. > >>>Section 2 (text above the box for "issue-require-targetnamspace") the > new text for targetNamespace is not very clean - maybe just for me. I > believe the resolution is to make defintion@targetNamespace required. All > children elements defined by this wsdl document belongs to this namespace. I > know we are talking about abstract model here, but if map to the wsdl > elements, the wording reads like wsdl1.2 requires that every individual > child elements, except type element, have a targetNamespace attribute. I > don't believe that's what we meant. > > Again, this may be just my own interpretation, other people may interpret > differently. To avoid future confusion, can we make the wording more > straight-forward? I think you're right, this needs rewording. NOT done. > >>>Section 2.3 paragraph 9 bullet 1 say "...one of the zero or more fault > messages MUST...". It sounds a little weird in the case of zero, should it > be changed to "...one of the listed fault messages MUST..."? Changed to: <replacement> The semantics are that when the input message is sent to the service, one of the following MUST happen: the output message is generated; or one of the fault messages listed is generated instead. It is an error for both an output message and a fault message to be generated in response to the same input message. </replacement> I think at some point we will need to get our terminology straight. For example, how do we name an endpoint? a service (like above), an endpoint (elsewhere in the spec), or an (ultimate) receiver (like in SOAP)? Is a message simply sent, or received and processed? Is an output message generated or sent back in response to a request? etc. > >>>Section 2.6 indicates that "A service description component MUST contain > one or more port descriptions", whereas section 1.2 (if we decide to keep > it) the grammar shows that service may have zero or more port - the symbol * > should be changed to + Done. > >>>Section 2.6 needs some clean up. There are some paragraphs duplicate or > even contradict with each other. For example, paragraph 4 and 6 both state > for port@name but makes different statements. paragraph 5 and 7 both refer > to "binding" property- seems paragraph 5 is wrong by saying "service" has a > binding property NOT done. > >>>section 4 optional @required for extensibility elements- If I remember > right, the WG agreed that the value of this attribute is "false". Should it > be explicitly reflected in the spec? Done. Added an extra sentence: <addition> Omitting this AII is defined as being semantically equivalent to including it with a value of false or 0. </addition> > TYPOS > > >>> section 2.2, paragraph 3: "a atomic entity" should be changed to "an > atomic entity" Done. > >>> section 2.2 paragraph 6: "if a type system other that XML Schema..." > should be changed to "if a type system other than XML Schema..." Done. > >>> section 3.2 paragraph 1: "WSDL uses the optional document elements..." > should be changed to "WSDL uses the optional documentation elements...". In > the last sentence of same paragraph, the font of "documentation" should be > changed to italic to be consistent with the rest of the spec Done both. Again, well spotted. > >>>section 3.3 paragraph 1, "by separationg the definitions ..." should be > changed to "by separating the definitions..." Couldn't find this one. NOT done. > I didn't dig into all sections for typos, you may just want to run a spell > check before publishing the spec. Done. Thanks again, Jean-Jacques.
Received on Tuesday, 2 July 2002 10:38:47 UTC