- From: Christopher B Ferris <christopher_ferris@us.ibm.com>
- Date: Thu, 27 Jun 2002 11:20:31 -0400
- To: "Jean-Jacques Moreau" <moreau@crf.canon.fr>
- Cc: Web Service Description <www-ws-desc@w3.org>, www-ws-desc-request@w3.org
Yes, I was only pointing out from a SOAP perspective. If WSDL doesn't
declare how it uses/doesn't use
xml:base, then requiring (or strong recommendation) that any URIs be
absolute is fine IMO.
Cheers,
Chris
"Jean-Jacques
Moreau" To: Christopher B Ferris/Waltham/IBM@IBMUS
<moreau@crf.canon cc: Web Service Description <www-ws-desc@w3.org>
.fr> Subject: Re: Issue 17: support for SOAP role attribute
Sent by:
www-ws-desc-reque
st@w3.org
06/27/2002 10:07
AM
Hi Chris, nice seeing you again! (I see you have your new name badge.)
I have been thinking about xml:base as well; IMO, whether a particular SOAP
message has a relative role, made absolute by
xml:base, or an absolute role, is implementation/SOAP specific. As far as
WSDL is concerned, it should not matter, the role
should be absolute; up to the SOAP sender to then decide to use xml:base or
not.
Does this make sense?
Jean-Jacques.
Christopher B Ferris wrote:
> Hmmm... SOAP allows use of xml:base[1] AII for purposes of establishing a
> base URI for any relative URI's carried in information items in, or
subordinate
> to, the SOAP envelope. It isn't clear to me how WSDL would capture this,
so that the base URI were applied to a SOAP
> envelope EII being created. In any event, URI values of any information
items that were ostensibly relative could be
> rendered absolute by means of prepending the base URI.
>
> Just a thought,
>
> Chris
>
> [1] http://www.w3.org/TR/2002/WD-soap12-part1-20020626/#reltoxml
>
>
> "Jean-Jacques
> Moreau" To: Martin Gudgin
<mgudgin@microsoft.com>
> <moreau@crf.canon cc: Jonathan Marsh
<jmarsh@microsoft.com>, Web Service Description
> .fr> <www-ws-desc@w3.org>
> Sent by: Subject: Re: Issue 17:
support for SOAP role attribute
> www-ws-desc-reque
> st@w3.org
>
>
> 06/27/2002 03:40
> AM
>
>
>
> Yes, the SOAP spec doesn't say. I think a relative URI does not make
sense
> in general, as the SOAP sender and receiver may have different assumption
> about the semantics carried by the URI.
>
> For example, from role="cacheManager", the receiver could infer
> role="http://nron.net/cacheManager" and the sender
> role="http://andersaine.gov/cacheManager". They may not understand each
> other well.
>
> On the other hand, if both agree before hand (via other means than WSDL),
> things will just be fine.
>
> So what about the following then (changes >>bracketed<<)?
>
> <proposed rev="3">
> The type of the role attribute information item is anyURI in the
> namespace named "http://www.w3.org/2001/XMLSchema". The value of
> the role attribute information item is a URI that names a role
> that a SOAP node can assume. >>It SHOULD NOT be
> a relative URI.<<
> </proposed>
>
> Jean-Jacques.
>
> Martin Gudgin wrote:
>
> > Good question. The SOAP 1.2 LC spec[1] doesn't say. An LC issue
perhaps?
> >
> > Gudge
> >
> > [1] http://www.w3.org/TR/soap12-part1/#soaprole
> >
> > -----Original Message-----
> > From: Jonathan Marsh
> > Sent: 26 June 2002 20:03
> > To: Martin Gudgin; Jean-Jacques Moreau
> > Cc: Web Service Description
> > Subject: RE: Issue 17: support for SOAP role attribute
> >
> > Can a role attribute contain a relative URI? xs:anyURI allows this,
> > unless we add a restriction in prose. If relative URIs are allowed,
can
> > we support the empty value case without stepping on the infoset [base
> > URI] property?
> >
> > > -----Original Message-----
> > > From: Martin Gudgin [mailto:mgudgin@microsoft.com]
> > > Sent: Wednesday, June 26, 2002 9:12 AM
> > > To: Jean-Jacques Moreau
> > > Cc: Web Service Description
> > > Subject: RE: Issue 17: support for SOAP role attribute
> > >
> > >
> > > Well, I think it sould probably just be 'role', as it appears on a
> > > soap:header element there is no need to qualify the attribute.
> > >
> > > Gudge
> > >
> > > -----Original Message-----
> > > From: Jean-Jacques Moreau [mailto:moreau@crf.canon.fr]
> > > Sent: 26 June 2002 16:25
> > > To: Martin Gudgin
> > > Cc: Web Service Description
> > > Subject: Re: Issue 17: support for SOAP role attribute
> > >
> > >
> > > This is much better, thanks. Actually, I think we should also change
> > > 'SOAP role' to 'soap:role'.
> > >
> > > Jean-Jacques.
> > >
> > > Martin Gudgin wrote:
> > >
> > > > Also suggest you merge the following two sentences;
> > > >
> > > > 'Omitting the SOAP role attribute information item is equivalent to
> > > > indicating value of
> > > > "http://www.w3.org/2001/12/soap-envelope/role/ultimateReceiver".
> > > > An empty value is equivalent to omitting the attribute completely,
> > > > i.e. targeting the SOAP header block to an ultimate SOAP receiver.'
> > > >
> > > > To read
> > > >
> > > > 'A SOAP role attribute information item that is either absent or
has
> >
> > > > an empty value is equivalent to indicating a value of
> > > > "http://www.w3.org/2001/12/soap-envelope/role/ultimateReceiver"
i.e.
> >
> > > > targeting the SOAP header block to an ultimate SOAP receiver.'
Received on Thursday, 27 June 2002 11:23:52 UTC