There is a new version of the specification up. Gisle and I are still
shaking out bugs so I haven't really encouraged anybody else to
implement it. Of course a third implementor will find more bugs but the
coordination costs get higher when we have to tell other people to
change their implementations because we found a bug.

Gisle is of course working on the Perl version. He's got pretty good
introspection stuff working. I'm working on a pseudo-Python version. I
say pseudo-Python because I don't have the time to write a full WSDL
parser in addition to everything else. Instead I am going to make the
base data structure and then allow other people to plug in their own
WSDL parsers. I am starting with IBM's WSDL4J parser (i.e. Jython!) as
an exercise in comparing what we have to what they have.

Hopefully I will have a beta version by Wednesday which should be a good
base for Rich's experimentations with ZSI. *BUT* there are two big holes
that we haven't worked out. One is the representation of complex types.
Gisle has something he's using but I haven't checked its applicability
to all languages and I certainly haven't turned it into a spec. 

The other is the binding between the WSDL reader and SOAP
implementations. Right now we are letting SOAP::Lite handle the
WSDL-based marshalling in Perl and IBM's WSIF handle it in Jython.
Really though, the WSDL library should control the SOAP marshalling --
it makes no sense to depend upon the SOAP implementation having its own,
independent implementation of WSDL. That we have two SOAP
implementations with partial WSDL support is just a quirk of history and
it won't work for ZSI, Tcl, etc.

Some of the issues to be handled here are:

 * schema-controlled marshalling and unmarshalling
 * binding registration
 * bindings "finding" their extension elements and attributes
 * representing bindings even for protocols that are not installed
 * exception reporting from bindings

If there is anybody out there that wants to tackle this latter issue,
which we haven't even started on, then give us a shout. Otherwise you'll
have to wait for us to get around to it. We are working up to it but we
are both busy and we have some complicated issues to work out around
complex types and schemas!

 Paul Prescod
_______________________________________________
Ws-dev mailing list
Ws-dev@[...].com
http://listserv.ActiveState.com/mailman/listinfo/ws-dev