SOAP Encoding - still a problem?

[Matthias Burbach <Ma...@web.de>]

Hi,

as you know there is a thing called SOAP Encoding (also known as SOAP
Section 5 Encoding in SOAP 1.1, now an optional addendum in SOAP 1.2 but
still present).

To me, SOAP Encoding is primarily an approach to support the
serialization of object graphs which are not strictly hierarchical, i.
e. identity of objects being referenced by multiple other objects will
be preserved.

However, SOAP Encoding has some flaws and created issues in the
interoperability of web services. This is, to my understanding, mainly
due to its serialization not being well founded on an XML schema.

Tim Ewald has analyzed this problem in an article on
http://msdn2.microsoft.com/en-us/library/ms995710.aspx

He also devised a way out of the misery in his bottom line which goes as
follows:
,Many people, myself included, believe that a shift away from SOAP
encoding is inevitable. The W3C XML Protocol Working Group's current
draft of the SOAP 1.2 specification makes support for SOAP encoding
optional (that is, a toolkit can claim SOAP 1.2 compliance without
supporting SOAP encoding), the WS-I Basic Profile Working Group's
current draft of its interoperability guidelines disallows the use of
SOAP encoding with SOAP 1.1, and the W3C Web Service Description Working
Group chose to drop support for encoding from their latest working draft
of the WSDL 1.2 specification.
It will take some time for toolkits to reflect these changes, first we
have to settle on a schema-friendly way to serialize graphs. Then
toolkits have to be updated. This will take some time, but it is worth
the wait. In the end, the Web service stack will be significantly easier
to implement and use.'

Tim's article dates back to 2002. Now, four years later I cannot see the
issue being resolved. Instead, all I can perceive almost everywhere in
web service publications is the recommendation to stay away from SOAP
Encoding (which is what the Basic Profile demands). However, I wonder if
this means developers are left to their own proprietary devices if they
have to preserve object graphs.

What do you think about this or what is it that I have missed in this
issue that makes me think there is a gap that needs to adressed?

Regards,
Matthias






---------------------------------------------------------------------
To unsubscribe, e-mail: axis-user-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-user-help@ws.apache.org