[
Lists Home |
Date Index |
Thread Index
]
On 29 Nov 2002 08:04:25 +0100, Eric van der Vlist <vdv@dyomedea.com> wrote:
>
> That's an interesting distinction which helps me understand why I feel
> uneasy with the W3C definition of Web services: it's like redefining
> "Web page" to be a page which has a URI and is described in XML!
See below for my exegesis on the term "XML" but that doesn't
seem like all that bad a definition of "web page" to me. A web page
MUST have a URI, right? OK, maybe the "described in" part should
refer to MIME types rather than XML. Even the ill-formed HTML out there
is "XML" in the sense that the "be liberal in what you accept" code in
the browsers essentially builds an XHTML Infoset (or DOM data model
if you prefer).
It would be an interesting exercise to define "web page" to the
satisfaction
of this list .... the definition of "web service" would probably "just"
tack on
the constraint that it must be relatively easy to write a program that does
what a human does on the Web as we know it: locate the URI of a resource
interest, get a representation via HTTP or some other transport /
transfer protocol [Hi Mark!], and process the results. IMHO, the most of
the disputes are about the details of technology and dogma, and "web
service" can be defined in a way that allows one to be implemented
with REST/XML, SOAP/WSDL, RDF/ontologies, and all sorts of interesting
combinations thereof.
>
> Also, it's missing your point by specifying that a Web service must be
> described as XML and not that this description should be published.
As anyone wishing to browse the www-ws-arch archives when insomnia
threatens will notice, this question comes up over and over. There is
a school of thought that says that "published" can mean some notes
scribbled on a piece of paper and sneakerneted from the producer of
a web service to the consumer. This point of view is particularly prevalent
among people who actually deploy web services within their
companies today -- people do what whatever it takes, formal or
informal, to communicate the structure of a SOAP message needed to invoke a
service and describe the format of the result.
>
> In other words, a Web resource described as XML in a document which
> stays on my laptop's hard disk meets the W3C definition while a Web
> resource which description would be published as WikiML documentation,
> Relax NG schema using the compact syntax and RDF/N3 does not.
Aha! I see the issue now ... That is MOST DEFINITELY not the intent of the
W3C WSA definition.I think we meant "XML" as the data model / Infoset (the
view enshrined in SOAP 1.2) [ducking the flames from Rich Salz and Tim Bray
:-) ]
and apparently people are reading the "XML" reference as insisting on XML
1.x syntax. Clearly the definition needs some wordsmithing!
BTW, there is a move afoot to ensure that WSDL 1.2 explicitly allows RELAX
NG
and RSS schemas as well as WXS schemas to define a web service. Someone
who knows more about that (Amy?) might want to fill this list in on its
status and prospects for inclusion in the eventual Recommendation.
|
