Stefan Tilkov wrote:
> On Dec 7, 2006, at 2:43 PM, Gregg Wonderly wrote:
>  > Stefan Tilkov wrote:
>  > > Bullshit. JAX-WS is as RPC-centric as the JAX-RPC/JAXM combination.
>  >
>  > Depending on what you transport and how the interface is described, RPC is
>  > equivalent to messaging. The standing issue for interoperability is a 
> common,
>  > known wire protocol. Maybe you can elaborate on the rift that you
>  > perceive?
> 
> RPC hides the fact that you're sending a message over a network, and
> tries (but fails) to create the illusion of a "remote method call".
> JAX-RPC basically treats the fact that you're sending XML messages
> over a net as an implementation detail. JAXM didn't, but
> unfortunately was (semi-officially) retired.
> 
> JAX-WS is 90% RPC, with a little messaging integrated - but any JAX-
> WS compliant implementation will have to support all of the RPC madness.

I guess this still doesn't help me understand.  From a programatic interface, 
in 
Java, if I can call a method and that method call results in using WS to send a 
SOAP message, that seems like a fine thing to me.  From the RMI programming 
model, it should also throw RemoteException or IOException or something to 
indicate a failure to connect and transfer the message.  I don't care about the 
wire protocol practically in my software.  Only for deployment/wiring would I 
actually need to know what was expected/provided by the service right?

Gregg Wonderly

Reply via email to