RE: Issue with soap-rpc:result

> With respect to typing, I don't see SOAP 1.1 as any more 
> dependent on XML schema than that.

But that seems pretty dependent to me.

Here's how I see it (as literally as possible with text):

Right now, we have this...

+-----------------+
| SOAP data model |
+-----------------+
   |          |
   | encoding |
   |          V
   |      +-------------------+
   |      |  XSD       other  |
   |      | simple      XSD   |
   |      | types     features|
   |      +-------------------+
   |          |
   V          V
+-----------------+
| XML-based tree  |
+-----------------+

But given that we're ending up with a serialized tree constrained by the
XML Infoset, couldn't we have this...

+-------------------+
|  SOAP data model  |
+-------------------+
   |          |
   | encoding |
   V          V
+-------------------+
|  XSD       other  |
| simple      XSD   |
| types     features|
+-------------------+
   |          |
   V          V
+-------------------+
|  XML-based tree   |
+-------------------+

This change would not require that SOAP nodes be any more XSD-aware then
they are now. Nor would it require validation or PSVI. It would allow us
to rely on a well-understood langauge for describing the messages that
are generated by the application of an encoding to the SOAP data model.
That in turn would open the door to use the current XSD-aware tools and
technologies, including validation and PSVI, when desirable. All that is
required to support this change is that the definition for SOAP encoding
be written in terms of XSD types instead of sample XML. And we would all
live in pease and harmony... :-)

Thanks,
Tim-

Received on Thursday, 7 February 2002 12:36:19 UTC