Re: encodingStyle

Paul,

SOAP 1.2 defines explicitly a graph data model and how graphs are
encoded in XML. SOAP 1.1 section 5 only does the latter without actually
defining the data model, but it's implied in there.

XML Schema - Structures - references a data model (the XML Infoset tree)
and allows to define (to constrain) some data structures on top of it.
WSDL uses XML Schema to communicate the restrictions on the data
structures. Apart from the straightforward "literal" interpretation of
the XML Schema, WSDL 1.1 mentioned (but did not define) an "encoded"
interpretation, which meant that the resulting instances didn't
necessarily have to validate against that schema using normal XML Schema
rules. The reason why we lost "encoded". We haven't lost encodingStyle
(as far as our issues list goes), we're still able to specify it (as a
hint) on message blocks.

The main graph-data-modle part that's missing from XML Schema is a typed
reference. I suspect adding typed references as a convention on how XML
Schema is to be created is doable, but I think it requires deep
understanding of obscure parts of the language. The result would likely
be incompatible with the S0AP 1.2 graph data model or its encoding.

If we want to support what SOAP 1.2 defines in this area, and if we want
to do it in a simple manner (see WS-Desc WG chapter, section 1.1), I
expect defining a simplistic schema language for the data model and
devising ways to use it in WSDL and maybe even in XML Schema itself is
the way to go. I'm working on a draft for this and while it's taken
overly long so far, I should have it done well before our next f2f.

It's possible that this schema language will fall into obscurity along
with SOAP 1.2 data model and encoding (which, AFAICS, is used very
little). In other words the market may decide XML Schema is sufficient
for XML data exchange. I'll have no problem with that.

As for non-XML serializations for SOAP - I can't see the relation
between this and encoding style issues.

Best regards,

                   Jacek Kopecky

                   Systinet Corporation
                   http://www.systinet.com/




On Fri, 2003-12-19 at 12:17, paul.downey@bt.com wrote:
> Following the call yesterday, i'm confused why we would need to invent a new
> schema language just to represent section 5 encoded messages when it's just
> a flag ATM. 
> 
> I'm also puzzled how other, non-XML serialisations for SOAP could be
> supported in the future - i'm guessing you'd need to invent a whole new 
> SOAP binding ?
> 
> Can someone please point me at the reasoning how we lost the encodingStyle
> on the SOAP binding ?
> 
> Paul

Received on Sunday, 4 January 2004 10:42:51 UTC