Ajith and Chathura,

If i want to write a DII kind of client  which doesn't use any static
stubs, then isn't it too much work to write the databinding code on my
own? I was expecting a version of Call.invoke() which can take
Object[]. Is this not in our wish list? What's our approach to support
RPC encoding?

- venkat


On 7/26/05, Chathura Herath <[EMAIL PROTECTED]> wrote:
>  
>  
> 
> Yup. The data binding is interfaced to the engine through the Message
> receiver that will also be generated. Serialization and deserialization is
> handled by the data binding framework. 
> 
>   
> 
> Chathura 
> 
>   
> 
>   
>  
>  
>  ________________________________
>  
> 
> From: Ajith Ranabahu [mailto:[EMAIL PROTECTED] 
>  Sent: Tuesday, July 26, 2005 2:05 PM
>  To: [email protected]; jayachandra
>  Subject: Re: [Axis2]Databinding Notes 
>  
> 
>   
> 
> Hi Jaya,
>  Well. You are quite right. Right now core has no such dependancy on the
> databinding. The idea is to incorporate databound objects when skeletons are
> code generated. Our plan is to generate the databound objects, skeletons and
> the relevant message receiver when skeletons are generated.
>  The marshaller and unmarshaller are built into the generated objects. To be
> exact we use our builders and the provided XML stream reader from the
> generated objects to marshall and unmarshall objects. 
>  
> 
> On 7/26/05, jayachandra <[EMAIL PROTECTED] > wrote: 
> 
> Ajith et al,
>  
>  In axis2 there is some databinding code present, but looks that only 
>  codegen of wsdl module is using it and not any other module like core
>  etc. seems like using it. The existing typemapping framework is
>  registering only the xml type and java type names in the map, and with
>  just that much information it becomes tough to actually output a java 
>  object out of a xml element and vice versa. Without marshaller and
>  unmarshallers registered for the types, can it be possible? If inside
>  the 'core' module code, we wish to data bind the parameters back and
>  forth (which would be a natural thing to do as we aim to support 
>  operations other than those which accept and output just OMElements,
>  correct me if I'm wrong here!) how can that be done?
>  
>  Thank you
>  Jaya
>  --
>  -- Jaya 
> 
> 
>  
>  
>  -- 
>  Ajith Ranabahu

Reply via email to