W3C home > Mailing lists > Public > www-tag@w3.org > October 2005

RE: EndpointRefs-47

From: David Orchard <dorchard@bea.com>
Date: Thu, 20 Oct 2005 09:46:21 -0700
Message-ID: <32D5845A745BFB429CBDBADA57CD41AF13CFCBCB@ussjex01.amer.bea.com>
To: "Henry S. Thompson" <ht@inf.ed.ac.uk>, <noah_mendelsohn@us.ibm.com>
Cc: <www-tag@w3.org>

It would make the association of EPRs to URIs dependent on a WSDL file,
perhaps similarly to making the association of an instance to a type
dependent upon a schema definition.  

I think the general idea is that an EPR minter that is using Reference
Parameters is saying "echo this back to me" and thus there isn't the
same need for generically self-describing QNames in URIs.  I think there
is a potential - though maybe small - problem that the receiver of a Ref
Param in a URI must have recreated the Ref Param QName before the
"dispatch on SOAP header" software is invoked and will be coupled to
that QName to URI binding.  

I fail to see the problem with coupling the RefP to URI mapping to a
WSDL file.  Here are the scenearios:

If an EPR minter uses some algorithm for mapping the RefP into the URI
and gives the EPR to a client, then the client either has the wsdl or
not.  
1) If it has the wsdl, then life is good.  
2) If it doesn't have the WSDL, then the client probably doesn't know
what the algorithm is (unless the EPR is adorned with the mapping
algorithm identifier but let's exclude that for now), then the client
will use the EPR via SOAP headers and won't use HTTP GET.  
3) If the client generates a URI using the algorithm that it knows about
and then gives that URI to some other software that doesn't know about
the algorithm, then presumably things are still good as the client has a
URI. 

The only failure case I can see is if somebody thinks that scenario #2,
that is an EPR with a Ref Param and No WSDL should not work when the Ref
Params are echoed as SOAP header blocks, and that hardly seems like a
failure case giving that's the entirety of the EPR world today.

Cheers,
Dave



> -----Original Message-----
> From: Henry S. Thompson [mailto:ht@inf.ed.ac.uk]
> Sent: Thursday, October 20, 2005 9:31 AM
> To: noah_mendelsohn@us.ibm.com
> Cc: David Orchard; www-tag@w3.org
> Subject: Re: EndpointRefs-47
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> noah_mendelsohn writes:
> 
> > Henry Thompson writes:
> >
> >> The suggestion made at the f2f, as I understood it, was to
> >> allow only the prefixes which were bound in the WSDL
> >> description identified by the primary URI, i.e. http://example.
> >> com/fabrikam/acct.
> >
> > This seems to make the association of EPRs to URIs dependent on a
> > particular WSDL file.  While it's true that many users of WSA will
also
> > use WSDL, I don't think that's required, and I don't think it should
be
> > required.
> 
> I am going to have to step back from this discussion until Tim BL or
> Dan C can recover more than I have been able to of our discussion in
> Edinburgh.  I thought the story which excited them at that time
> depended on there being a reliably-shared source of namespace
> declarations == prefix bindings at the end of the primary URI.  If
> that's not plausible, I agree the story fails.
> 
> Given that the original discussion was in the context of WSRF, is
> it perhaps the case that WSRF does _require_ a WSDL document, or else
> requires some _other_ document which equally reliably provides the
> relevant declaration/binding.
> 
> noah_mendelsohn writes:
> 
> > You wind up pretty close to a situation where the only actor that
> > really understands the URI is the server that minted it, in which
> > case it's not clear why globally unique naming of the parameters is
> > of great value at all.  If the server knows the WSDL, then
> > presumably it can use:
> >
> >     http://example.com/fabrikam/acct?parm1=412
> >
> > since it knows that the WSDL calls for only one parameter anyway.
> >
> > I presume that the WSA folks are using globally unique parameter
names
> for
> > a reason, and I'm not yet convinced that grounding the URI mappings
in a
> > WSDL file is appropriate.
> 
> My understanding was that problem this approach solved was when a WSDL
> document imported some widely-used sharable artefact which defined
> e.g. an
{http://www.example.org/WidelyUsed/PhysicsParams/}ExperimentCode
> reference param. . .
> 
> But again, I'm hoping the originally enthusiastic participants can
> reconstruct the basis for their enthusiasm -- I'm just trying to
> provoke them!
> 
> ht
> - --
>  Henry S. Thompson, HCRC Language Technology Group, University of
> Edinburgh
>                      Half-time member of W3C Team
>     2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131
650-4440
>             Fax: (44) 131 650-4587, e-mail: ht@inf.ed.ac.uk
>                    URL: http://www.ltg.ed.ac.uk/~ht/
> [mail really from me _always_ has this .sig -- mail without it is
forged
> spam]
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.6 (GNU/Linux)
> 
> iD8DBQFDV8YrkjnJixAXWBoRAoEcAJ9cxnVi2QrZQynC7afwDkmZn71k4QCfcQ8U
> 3RjT7Cs4JtCOwo/A0AyMVkw=
> =5mqD
> -----END PGP SIGNATURE-----
Received on Thursday, 20 October 2005 16:46:50 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:56:09 UTC