RE: Abstract messages [Was: Multi-Party Binding Scenario]

Jon Dart sent:


> +1
> 
> I don't think I said anything different from this - IMO 
> constructing an 
> XML-based binding for something like ASN.1 is practically 
> difficult, and 

not any more. There are a set of recent standards from the ASN.1 group
that allow XML representation of
types defined in ASN.1 and, by conversion of XML schemas to ASN.1
definitions, the ASN.1 encoding of
XML defined types.   See http://asn1.elibel.tm.fr/xml/ for details (and
a schema -> asn.1 converter
that makes it a lot easier to see what's what).

One (deliberate) effect of this work is that XML schema-defined types
can be transmitted using ASN.1
Packed Encoding Rules, which are highly compressed (and have a canonical
encoding). 

There is no difference between ASN.1 and XML - they are just alternative
notations of the same thing !

Peter



> even if it were possible to do this without relying on WSDL, such a 
> binding would be out of scope for this WG. I think there is 
> still some 
> support for preserving this as a possibility, but I don't personally 
> favor that view. Even exotic message formats can in principle be 
> transformed to XML - and doing so is easier IMO than trying 
> to deal with 
> them somehow in their original form.
> 
> How tightly tied WSDL is to the choreography definition can still be 
> discussed, e.g. whether it is directly embedded or indirectly 
> referenced, and if the latter, how much and what kind of 
> indirection is 
> desirable.
> 
> --Jon
> 
> Jim Webber wrote:
> > Jon, Tony,
> > 
> > Furthermore given the name (and charter?) of this group I would 
> > suggest that tying its work to Web services (and perhaps 
> the work of 
> > the WSAG) would be the most sensible thing to do. While the 
> notion of 
> > an abstract and bindable choreography standard does have a certain 
> > undeniable appeal, I think this group should really stick to a 
> > specific "binding" if you like (and XML Schema plus WSDL 
> works for me 
> > because you can extend it nicely), and that should be for 
> Web services 
> > (pick the definition that best suits you, but I guess things like 
> > ASN.1 are out of scope).
> > 
> > Jim
> > 
> > 
> > 
> > 
> 
> 
> 
> 

Received on Friday, 4 April 2003 08:29:27 UTC