RE: [wsrp-webservice] optional desiredLocales parameter

[Andre Kramer <andre.kramer@eu.citrix.com>]

Title:

Hi David,

 

You can certainly edit out the "namespace="http://www.w3.org/2001/XMLSchema"/" constraint before you compile the wsdl. Do you currently need to make any other edit to make 0.93 work for JAX-RPC?

 

We need to be able to get at the other xsd basic types (such as xsd:int) as well as strings, so we all must be able to walk a XML node graph (as types other than xsd:string) will be found in the "any". So I don't think it unreasonable to work around this issue as well. [I assume you are able to still get at the stringValue when JAXRPC de-serializer gives up)? I only suggest this as adding a ComplexValue element would change the spec as well as the wsdl.

 

regards,

Andre

 

PS. I passed the Oracle offices in Reading, Tuesday. Would have dropped in had I known which building you are in ...

-----Original Message-----
From: David Ward [mailto:david.ward@oracle.com]
Sent: 10 April 2003 12:38
To: Andre Kramer
Cc: wsrp-webservice@lists.oasis-open.org
Subject: Re: [wsrp-webservice] optional desiredLocales parameter

Hi. I assume this is the right list to send WSDL issues to!

I have noticed that in the 0.93 WSDL we still have "<any namespace="http://www.w3.org/2001/XMLSchema"/>" in the ModelTypes definition. I have requested on many occasions now for this to be changed to "<any namespace="##other"/>" because our JAX-RPC stack can't seem to cope with a fixed namespace and it causes the entire ServiceDescription (not just the ModelTypes) to be serialized as an untyped SOAPElement! It is now acceptable that to use the WSRP WSDL with certain SOAP stacks, you may be forced to make edits to it?

While I'm on the same subject, it would be really useful if you could wrap up the "<any namespace="##other" minOccurs="0" maxOccurs="unbounded"/>" in the Property type up in its own named element, so that we can still get at stringValue without having to deal with an untyped SOAP element (this is just a minor inconvenience). What's wrong with having stringValue and complexValue?

Thanks

David

Andre Kramer wrote:

After yesterdays call I decided to go ahead and give a "minOccurs='0'
desiredLocales" a quick test .NET <--> Axis 1.0.

using: 
<element name="desiredLocales" type="xsd:string" minOccurs="0"
maxOccurs="unbounded" />

instead of (3 uses):
<element name="desiredLocales" type="xsd:string" nillable="true"
maxOccurs="unbounded" />

results (same for all .NET <--> Axis combinations):

sending null for desiredLocales - received desiredLocales == null
sending new String[0] for desiredLocales - received desiredLocales == null
sending new String[] { "ga" } - received desiredLocales.length == 1
desiredLocales[0].equals("ga"}
sending new String[] { "ga", "de" } - you get the picture, I'm sure.

So (if Axis is used with the --noWrapped flag) having an optional array
param is ok (note: the optional array is always in the last param position -
for above test & our wsdl).

regards,
Andre
  

--

---

David Ward
Principal Software Engineer
Oracle Portal

Oracle European Development Centre
520 Oracle Parkway
Thames Valley Park
Reading
Berkshire RG6 1RA
UK

Email:	david.ward@oracle.com
Tel:	+44 118 924 5079
Fax:	+44 118 924 5005