OK, that makes sense.  You don't have a wrapper elem to find the operation
from.

Do you see any harm in just doing the same (generating a similar soapAction)
for a wrapped-style operation?

The background here is that I'm looking at some furthering the AXIOM->JAXB
optimizations we started.  It seems that, without the soapAction present, we
have to build enough of the OMElement to enable the dispatch, but at a point
which defeats the optimization I'm trying to do.

Scott



On Mon, Oct 13, 2008 at 12:42 PM, Raymond Feng <[EMAIL PROTECTED]> wrote:

> Hi,
>
> I added the code to make sure the Axis2 can figure out the operation for
> BARE style based on SOAP action.
>
> Thanks,
> Raymond
>
> From: Scott Kurz
> Sent: Monday, October 13, 2008 8:59 AM
> To: [email protected]
> Subject: runtime Java2WSDL and soapAction
>
>
>
> I noticed, in file:
>
> java/sca/modules/binding-ws-wsdlgen/src/main/java/org/apache/tuscany/sca/binding/ws/wsdlgen/Interface2WSDLGenerator.java
>
> that we have the following code:
>
>           String action = ((JavaOperation)op).getAction();
>           if ((action == null || "".equals(action)) && !op.isWrapperStyle()
> && op.getWrapper() == null) {
>               // Bare style
>               action = "urn:" + op.getName();
>           }
>
> i.e. we calculate a default soapAction for the WSDL we're generating if
> this is BARE style.
>
> I think this is useful, in spite of the fact that JSR-181 says that the
> default soapAction should be unset.
>
> Just wondering if anyone knew the rationale behind making a special case
> for BARE style... then maybe we could discuss
> whether it might make sense to do it for WRAPPED as well.
>
> Scott
>

Reply via email to