W3C home > Mailing lists > Public > www-ws-desc@w3.org > April 2002

Re: issue: support cross references within a WSDL file using ncnames?

From: Roberto Chinnici <roberto.chinnici@sun.com>
Date: Wed, 17 Apr 2002 10:32:10 -0700
Message-ID: <3CBDB19A.22766CAF@sun.com>
To: Martin Gudgin <martin.gudgin@btconnect.com>
CC: www-ws-desc@w3.org
+1 to eliminate duplicate operation names in interfaces.

If the publisher of a WSDL document wants to provide a hint on
how operation names should be mapped to (possibly overloaded)
method names, it can use operation metadata to do that. Naturally,
clients are free to disregard the hint and use their preferred language-
and environment-specific mapping instead.

Roberto

--
Roberto Chinnici
Java and XML Software
Sun Microsystems, Inc.


Martin Gudgin wrote:

> There are no contraints on operation names beyond the fact that they must be
> of type NMTOKEN. Multiple operations in a given portType may have the same
> name, which means we must look at the names of the input/output/fault pieces
> to figure out the mapping. This does not seem particularly good to me. I'm
> not entirely certain I understand why duplicate operation names are allowed.
> Why not make them unique? I don't really buy the method overloading argument
> as I don't see why operation names need to map directly to method names.
> Method name seems to be an RPCism to me, why is it part of the definition of
> an operation?
>
> Gudge
>
> ----- Original Message -----
> From: "Martin Gudgin" <martin.gudgin@btconnect.com>
> To: "Martin Gudgin" <martin.gudgin@btconnect.com>; "Francisco Curbera"
> <curbera@us.ibm.com>
> Cc: <www-ws-desc@w3.org>
> Sent: Wednesday, April 17, 2002 11:47 AM
> Subject: Re: issue: support cross references within a WSDL file using
> ncnames?
>
> > I also think that given the operation names have to be unique in a given
> > document ( aside: should that really be 'targetNamespace' ) then we really
> > do have another symbol space for operations alongside those for messages,
> > portTypes and bindings.
> >
> > Gudge
> >
> > ----- Original Message -----
> > From: "Martin Gudgin" <martin.gudgin@btconnect.com>
> > To: "Francisco Curbera" <curbera@us.ibm.com>
> > Cc: <www-ws-desc@w3.org>
> > Sent: Wednesday, April 17, 2002 8:50 AM
> > Subject: Re: issue: support cross references within a WSDL file using
> > ncnames?
> >
> >
> > >
> > > ----- Original Message -----
> > > From: "Francisco Curbera" <curbera@us.ibm.com>
> > > To: "Martin Gudgin" <martin.gudgin@btconnect.com>
> > > Cc: <www-ws-desc@w3.org>
> > > Sent: Tuesday, April 16, 2002 8:51 PM
> > > Subject: Re: issue: support cross references within a WSDL file using
> > > ncnames?
> > >
> > >
> > > >
> > > > Martin,
> > > >
> > > > There is no requirement that operation names be unique within a
> > > > targetNamespace.
> > >
> > > I know.
> > >
> > > > That means that you can only specify an operation *within*
> > > > a portType,
> > >
> > > I know that too.
> > >
> > > > and that an NCName is sufficient for that.
> > > > What would be then
> > > > the benefit  of namespace qualifying the operation name?
> > >
> > > That all references in WSDL would be consistent. Consistency is good.
> > >
> > > Gudge
> > >
> > > >
> > > > Paco
> > > >
> > > >
> > > >
> > > > "Martin Gudgin" <martin.gudgin@btconnect.com>@w3.org on 04/16/2002
> > > 08:40:27
> > > > AM
> > > >
> > > > Sent by:    www-ws-desc-request@w3.org
> > > >
> > > >
> > > > To:    <www-ws-desc@w3.org>
> > > > cc:
> > > > Subject:    Re: issue: support cross references within a WSDL file
> using
> > > >        ncnames?
> > > >
> > > >
> > > >
> > > > Ahhh, if only default namespace decls didn't exist, we wouldn't have
> > this
> > > > confusion...
> > > >
> > > > Names in the XML 1.0 + Namespaces in XML world are either qualified or
> > > > unqualified. Whether a name is qualified or unqualified is largely
> > > > orthogonal to whether it is prefixed or unprefixed. I say largely
> > because
> > > a
> > > > prefixed name is always a qualified name.
> > > >
> > > > All references in WSDL SHOULD be QNames. Unfortunately certain
> > references
> > > > in
> > > > WSDL are not QNames ( e.g. mapping between operations in a binding and
> > > > operations in a portType ). I know that the reference from binding to
> > the
> > > > portType is by QName, it just seems weird that the mapping of
> operations
> > > is
> > > > by local name.
> > > >
> > > > Gudge
> > > >
> > > >
> > > >
> > > > ----- Original Message -----
> > > > From: "Simon Fell" <soap@zaks.demon.co.uk>
> > > > To: <www-ws-desc@w3.org>
> > > > Sent: Tuesday, April 16, 2002 1:55 AM
> > > > Subject: Re: issue: support cross references within a WSDL file using
> > > > ncnames?
> > > >
> > > >
> > > > > On Mon, 15 Apr 2002 06:56:22 -0400, in soap you wrote:
> > > > >
> > > > > >>"Simon Fell" <soap@zaks.demon.co.uk> writes:
> > > > > >> >Yes, the references across NSs always have to be QNames. I
> believe
> > > > > >> >telling whether a name is a QName or an NCName is simple- just
> see
> > > > > >> >whether there's a colon or not.
> > > > > >>
> > > > > >> but an unprefixed value, is still a valid QName.
> > > > > >
> > > > > >Isn't it a QName with a null namespace URI? Is there a difference
> > > > > >between that and an NCName??
> > > > >
> > > > > No, its a QName with a null prefix. the namespace URI will depend on
> > > > > whatever the default namespace is at that scope.
> > > > >
> > > > > >> It sounds to me like its just making it more complex, not less.
> > Stick
> > > > > >> with QNames and make it clearer in the prose.
> > > > > >
> > > > > >That would be my own personal preference too!
> > > > > >
> > > > > >Sanjiva.
> > > > >
> > > >
> > > >
> > > >
> > > >
> > >
> >
Received on Wednesday, 17 April 2002 13:36:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:19 GMT