Hi Asankha,
There are a couple of reasons for this code..
1. During interoperability testing of the WS-A WSDL Binding Doc (in
the W3C), other participants indicated that they though that that
fault must be thrown if the received action is incorrect.
2. (Separately) During interoperability testing with MS endpoints, the
MS endpoint would reject a message that had an incorrect action,
whereas when the client and endpoint is axis2, everything is fine.
This left a general perception that there is an interoperability
problem between Axis2 and MS whereas the truth is the the client has
been incorrectly developed. I'd like client authors to find the
problem when working in the all Axis2 environment rather than it
appearing as a problem later on.

I suspect we could (and perhaps will) disagree on whether we should be
giving the user rope  and a wobbly box to stand on to hang themselves
but my preference is not to.

I do, however, (as with the mustUnderstand case) agree that to use
Axis2 in an ESB, up-front validation may not be desirable.

There are therefore 3 routes forward I'd like you to consider
1. Synapse deploys its SynapseDispatcher between the
AddressingBasedDispatcher and the AddressingValidationHandler
2. Synapse deploys a custom module.xml which does not include the
AddressingValidationHandler
3. We allow configuration of the AddressingValidationHandler through a parameter

I'm happy with any of these, what do you think?
David

On 19/03/07, Asankha C. Perera <[EMAIL PROTECTED]> wrote:
David

I assume you are familiar with this code, and I do not want to make a
fix without knowing the reasons behind.. hope you could reply to me
whether this is a defect, or the way it was designed. If its a defect I
could file a JIRA and probably fix it..

thanks
asankha

Asankha C. Perera wrote:
> Hi
>
> Why does the AddressingValidationHandler throw an ActionNotSupported
> fault if a message contained addressing, but fails to dispatch using
> addressing? In Synapse, we place a custom SynapseDispatcher to grab
> any messages that have not been dispatched by any of the other
> dispatchers, and if the AddressingBasedDispatcher fails, we know how
> to handle it. Also this worked until Axis2 1.1.1, but now it is broken
>
> asankha
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]




--
David Illsley - IBM Web Services Development

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to