RE: Review of WSDL core for WSA

*	Section 3.3 has the intriguing title "Describing Messages that
Refer to Services and Endpoints." which seems at least statistically
similar to the EPR section of the WSA core, and it even talks about the
{address} property of endpoints.  However, on closer inspection this is
just about tagging schema components and the like with the names of
interfaces and bindings that reference them.  I don't believe this
affects us, and I don't recall it coming up in our discussions.

This is the section I highlighted in my mail requesting review [1].
Specifically, does the WS-A WG have any concerns about WSDL 2.0
recommending the use of these annotations on types derived from
wsa:EndpointReference?

 


[1]
http://lists.w3.org/Archives/Public/public-ws-addressing/2005Aug/0070.ht
ml

 

________________________________

From: public-ws-addressing-request@w3.org
[mailto:public-ws-addressing-request@w3.org] On Behalf Of David Hull
Sent: Saturday, September 10, 2005 10:08 PM
To: public-ws-addressing@w3.org
Subject: Review of WSDL core for WSA

 

I've gone through the WSDL core LC2 draft from top to bottom a few times
now, and also had a fresh look at our WSDL binding.  While I do have
several comments and questions about the core on my own behalf, which
I'll be writing up as time permits, I don't believe there's too much to
say from a WSA perspective.

As far as I can tell, WSA and WSDL intersect mainly in the area of MEPs.
EPRs can carry WSDL descriptions as metadata, but they can carry
anything at all as metadata, so it's hard to see what restrictions this
would entail.  MEPs, on the other hand, are dealt with mainly as
adjuncts, not in the core.  As long as the WSDL can support MEPs beyond
the core set, and I believe it does so by design, we should be OK there.

Long story short, only two areas stood out on first reading:

*	Section 3.3 has the intriguing title "Describing Messages that
Refer to Services and Endpoints." which seems at least statistically
similar to the EPR section of the WSA core, and it even talks about the
{address} property of endpoints.  However, on closer inspection this is
just about tagging schema components and the like with the names of
interfaces and bindings that reference them.  I don't believe this
affects us, and I don't recall it coming up in our discussions.
*	Section 6.1.1, "Mandatory extensions" is about the wsdl:required
attribute.  This, of course, has been the subject of quite a bit of
discussion concerning how to use this attribute in conjunction with
wsa:UsingAddressing.  Reading through this gave me a bit more insight
into that discussion, but as far as comments to WSDL, I don't really see
any.  From our point of view, wsdl:required is a "fact on the ground,"
and unless we feel that wsa:UsingAddressing can't be made to work
without changes to it, I don't think we have anything  to add.  I
believe our woes have been more along the lines of trying to understand
which of several possible readings of wsdl:required to standardize on,
and that's our problem, not WSDL's.  WSDL isn't preventing us from
defining the semantics we need.

With this in mind, I believe I have discharged my action item for WSDL
review to the best of my ability.  If anyone else has any major
concerns, please speak up.