W3C home > Mailing lists > Public > www-ws-desc@w3.org > March 2003

Re: Remove binding? ( was RE: Proposed renamings)

From: Sanjiva Weerawarana <sanjiva@watson.ibm.com>
Date: Mon, 3 Mar 2003 20:51:20 +0600
Message-ID: <01d701c2e194$5c4e9b20$6f00a8c0@lankabook2>
To: "Martin Gudgin" <mgudgin@microsoft.com>, "Jean-Jacques Moreau" <jean-jacques.moreau@crf.canon.fr>, "Philippe Le Hegaret" <plh@w3.org>
Cc: <www-ws-desc@w3.org>

Can you post an example of this please?

In any case, this is a major change from the fundamental language. I'm
under the impression that Jonathan wants to close the part 1 doc for
the most part this week. Jonathan, is that not the case?

I agree with what Jack said too - interfaces and bindings are totally
different beasts. Just because something seems like a good way to
shortcut the syntax or the component model that doesn't mean its the
right thing to do. PortTypes (interfaces) are about describing what
a service can do at an abstract level and bindings allow one to specify
how that functionality is made available. Combining the two things
via an inheritance relationship (where none exists!) is just
non-sensical to me.

Sanjiva.

----- Original Message -----
From: "Martin Gudgin" <mgudgin@microsoft.com>
To: "Jean-Jacques Moreau" <jean-jacques.moreau@crf.canon.fr>; "Philippe Le
Hegaret" <plh@w3.org>
Cc: <www-ws-desc@w3.org>
Sent: Friday, February 28, 2003 4:38 PM
Subject: Remove binding? ( was RE: Proposed renamings)


>
> So, at the component level, binding is almost identical to port type; the
only difference is that you can specify multiple bases for a port type, but
only one 'base' for a binding. So, why don't we get rid of binding
completely and use port type inheritance to layer in the binding info? For
example, given port types
>
> A
> |
> B
>
> where A and B are port types as they stand today. Then
>
> B
> |
> C
>
> where C contains binding extensions related to message serialization and
says "we'll serialize the messages using a SOAP Envelope"
>
> and then
>
>  C
> | |
> D E
>
> where D and E contain other binding extensions related to actual
transportation of bits and say "we'll send the bits over HTTP" (D) or "We'll
send the bits over TCP" (E)
>
> This seems nice and flexible to me, it's clean at the component level, the
syntax would be the same level of complexity as we have today and it gives
us a way to do layered bindings, which have been mentioned by several
people. People can build using as many or as few layers as they like.
>
> Comments, flames etc. to the usual address
>
> Gudge
>
> ________________________________
>
> From: Jean-Jacques Moreau [mailto:jean-jacques.moreau@crf.canon.fr]
> Sent: Fri 28/02/2003 08:57
> To: Philippe Le Hegaret
> Cc: www-ws-desc@w3.org
> Subject: Re: Proposed renamings
>
>
>
>
> Is there any keyword other than <interfaceBinding> that we could
> use to help alleviate the confusion between a WSDL binding and a
> SOAP binding?
>
> A SOAP binding specifies how to (physically) transport a SOAP
> message over an undertlying protocol. A WSDL binding specifies
> how abstract messages are serialized and what protocol is used.
> It does not in itself indicate how the messages are physically
> transported. The terms are different, but complementary, and
> apply to different layers. I have seen so much confusion between
> the two, even on W3C mailing lists, that I would be keen to name
> them differently.
>
> Jean-Jacques.
>
> Philippe Le Hegaret wrote:
> > [I thought I sent these yesterday but don't see it in the archives, so
> > sending it again]
> >
> > I've got an action item to start a proposal on renaming elements and/or
> > attributes in WSDL 1.2. This proposal is based on the latest WSDL 1.2
> > drafts and the requirements document. I'll keep track of sub-sequa=ente
> > proposals
> >
> > - portType
> >  The requirements document has the following:
> >  [[
> >  Interface (AKA Port Type)
> >   [Definition: A logical grouping of operations. An Interface represents
> >   an abstract Web Service type, independent of transmission protocol and
> >   data format.]
> >  ]]
> >  http://www.w3.org/TR/2002/WD-ws-desc-reqs-20021028/#normDefs
> >
> > In 2.4.2 XML Representation of Port Type Component:
> > http://www.w3.org/TR/2003/WD-wsdl12-20030124/#PortType_XMLRep
> >
> >  A [local name] of portType
> >
> >  would read
> >
> >  A [local name] of interface
> >
> > -  EndPoint (AKA Port)
> >  The requirements document has the following:
> >  [[
> >  EndPoint (AKA Port)
> >   [Definition: An association between a fully-specified InterfaceBinding
> >   and a network address, specified by a URI [IETF RFC 2396], that may be
> >   used to communicate with an instance of a Web Service. An EndPoint
> >   indicates a specific location for accessing a Web Service using a
> >   specific protocol and data format.]
> >  ]]
> >  http://www.w3.org/TR/2003/WD-wsdl12-20030124/#Port_XMLRep
> >
> > In 2.11.2 XML Representation of Port Component
> >
> >  A [local name] of port
> >
> >  would read
> >
> >  A [local name] of endPoint
> >
> >
> > We may also do the following but, if we don't, I'll recommend changing
> > the glossary of our requirements document.
> >
> > - binding
> >  The requirements document has the following:
> >  [[
> >  InterfaceBinding
> >   [Definition: An association between an Interface, a concrete protocol
> >   and/or a data format. An InterfaceBinding specifies the protocol
> >   and/or data format to be used in transmitting Messages defined by the
> >   associated Interface.]
> >  ]]
> >  http://www.w3.org/TR/2002/WD-ws-desc-reqs-20021028/#normDefs
> >
> > In 2.7.2 XML Representation of Binding Component
> > In http://www.w3.org/TR/2003/WD-wsdl12-20030124/#Binding_XMLRep
> >
> >  A [local name] of binding
> >
> >  would read
> >
> >  A [local name] of interfaceBinding
> >
> >
> >
>
>
Received on Monday, 3 March 2003 09:54:34 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:23 GMT