RE: [wsrp-webservice] call

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

Title: [wsrp-webservice] call

While the warning is a concern, it appears messages deserialize ok for .NET.  So this seems a useful implementation technique for .NET. We should check it allows type safe de-serialization with the other stacks also. But I would not use this approach in our official published wsdl (only as an implementation trick).

 

Also,  we need to decide what the ModelDescription for our string array properties should be.

 

a) multiple occurrences of property with the same name: Its PropertyDescription.type is not really "xsd:string" but I can't think of a good QName to use.

 

b) use of "extra" schema: PropertyDescription.type == "urn:oasis:nam:tc:wsrp:v1:extra". But would we carry the extra schema in ModelDescription.ModelTypes?

 

c) modify Property to add an in-line stringArray from our namespace: PropertyDescription.type == "urn:oasis:name:tc:wsrp:v1:types:StringArray".

 

I realise (c) requires a version rev but is it not the most natural way to support string[] properties?

 

    -- Andre

-----Original Message-----
From: Andre Kramer [mailto:andre.kramer@eu.citrix.com]
Sent: 12 November 2003 16:41
To: wsrp-webservice@lists.oasis-open.org
Subject: RE: [wsrp-webservice] call

I tried out a second extra .xsd schema (attached) as follows:

 

added import to wsrp_v1_types.xsd:

 

    <import namespace="urn:oasis:names:tc:wsrp:v1:extra" schemaLocation="wsrp_v1_extra.xsd" />

 

modified our Property type to add "extra:StringArray" element:

 

<complexType name="Property">

<sequence>

<!-- Would prefr this to be a choice ... Axis and JAX-RPC failed to handle that -->

<element name="stringValue" type="xsd:string" minOccurs="0" />

<element ref="extra:StringArray" minOccurs="0" />

<any namespace="##other" minOccurs="0" maxOccurs="unbounded" />

<!-- end prefer this to be a choice -->

</sequence>

<attribute name="name" type="xsd:string" use="required" />

<attribute ref="xml:lang" />

</complexType>

.NET gave me a warning:

Schema validation warning: The content model must be deterministic. Wildcard declarations along with a local element declaration causes content model to become ambigous.

Seemed to code correctly though ...

regards,

Andre

-----Original Message-----
From: Andre Kramer [mailto:andre.kramer@eu.citrix.com]
Sent: 12 November 2003 15:42
To: wsrp-webservice@lists.oasis-open.org
Subject: [wsrp-webservice] call

Calling in: Rich, Richard, Andre

- discussed status of attachments work for 1.1.
        - We now have some data on gains to expect from use of attachments and an opportunity to test both SwA and DIME mechanisms. However, standardization & best practice lags, and the group felt that we should be following behind others. In particular, we should await type safe (wsdl) means to specify attachments. Likely to be post 1.1.

        - new proposals seem to be leaning towards an abstract model for attaching non-DOM data to an Infoset and we should explore which of our protocol elements could benefit. Action: Andre to identify candidates for next call.

- we discussed Richard's proposal to publish a non-normative .xsd (XML Schema) file on our TC Web Site. This would contain StringArray and NamedStringArray types and elements (unfortunately we can't selectively import the types form the existing schema for re-use?) but defined in another namespace. We would not import this extra schema into our wsdl but would expect specific consumers and producers to do so (for multi-valued JSR 168 string property, for example).

Action: Andre to raise in tomorrow's TC call.

        -- Andre