RE: Back to Requirements (was RE: Web Service Definition [Was "So me T houghts ..."])

> -----Original Message-----
> From: Christopher Ferris [mailto:chris.ferris@sun.com]
> Sent: Tuesday, March 05, 2002 3:39 PM
> To: Mark Baker
> Cc: www-ws-arch@w3.org
> Subject: Re: Back to Requirements (was RE: Web Service Definition [Was
> "Some T houghts ..."])
> 

> I used the term "XML Infoset" which is not the same as
> XML1.0 anglebrackets and so does not constrain the "what"
> which amounts to some mechanism by which an XML Infoset
> is "expressed" (serialized).

Does it constrain the definition enough to be useful?  Arbitrary
HTML can be expressed in the InfoSet once the parser has done its
voodoo to fix the non-well-formedness allowed by HTML syntax.
I'm not sure about this, but doesn't the SOAP encoding allow just about
any programming data structure to be mapped onto XML syntax, and 
hence the infoset?  If any document and any data can be represented in
the InfoSet, why bother mentioning it?  (Maybe I'm missing something!)

I still see one of two ways out: we either say that we are only defining
XML Web Services, and they MUST serve up data in XML syntax (and all the
things that don't can be "web services" but not "XML web services", and 
thus of limited interest to us) ... or we say that we are defining "web
services" broadly but are constrained to build a reference architecture
that employs XML technology wherever appropriate in the definition,
invocation, results, security policy, directory services, etc.  I
for one prefer the latter.  Our charter can constrain what WE do, but
it can't constrain how the WORLD defines "web services".

Received on Tuesday, 5 March 2002 16:03:28 UTC