Re: [wsrf] ResourceProperty comments from HP

Brian, comments on some of your review comments.

Regards,
Ian Robinson
STSM, WebSphere Messaging and Transactions Architect
IBM Hursley Lab, UK


"Murray, Bryan P." <> wrote on 31/08/2005 18:25:40:


> 2. Introduction talks about standardizing the means by which properties
> are defined. We really are not specifying how a schema is defined, just
> that there be a link to it from the WSDL doc - so we are not
> standardizing the means the properties are defined.
The introduction (line 102) says: "This specification standardizes the
means by which the definition of the properties of a WS-Resource may be
declared as part of the Web service interface. " This does not seem
unreasonable or incorrect

> 4. I would remove the non-goal section, partly because we are moving
> that direction, partly because it is not very relevant.
This is useful and reflects a statement to the same effect in the charter.
The purpose of this statement is to make clear that properties referenced
in WSRF-RP messages are identified by QName. WS-RF is not competing with
XQuery Update.

> 6. Line 163: this paragraph should be removed because there are no
> references to infoset. In fact every use of brackets indicates one of
> the references in the reference section and not an infoset property.
> Also, the "[XML Infoset]" reference should be "[XML-Infoset]" to match
> the reference section.
This comment applies equally to all the WS-RF specs


> 7. The SOAP 1.1 namespace is used. Since we only show the wsa:Action
> header with no mustUnderstand, there is no need to be restricted to SOAP
> 1.1. Lets do what WSN did and use the prefix 's' to mean both versions
> of SOAP.
We specifically resolved to use SOAP 1.1 for the examples in issue 99.
These are non-normative but we nevertheless declared the namespace in
section 1.3.

> 10. I think we could remove the interface field from all of the
> wsa:Action definitions since that interface will never be realized.
The wsa:Action definitions are deliberately chosen to follow the defaults
specified in the WS-Addressing WSDL Binding spec:
[target namespace][delimiter][port type name][delimiter][input|output name]
The advantage of using this default is that it makes it unnecessary for the
service provider to explicitly specify wsa:Action attributes in their WSDL.

> 11. Line 389: The spec says "... comprise all of the resource property
> values ...". It seems to me that the contents comprise only that portion
> of the XML representation of the resource that the resource is willing
> to give you. For security reasons it may not give you everything. The
> client should be ready to take what it can get and not necessarily
> expect to get everything. Also, this statement is repeated for every
> message exchange, wouldn't it be better to just state it once and have
> it apply to all message exchanges.
Can you propose a concrete statement that the specification should make in
place of this?