Asankha

As a post 1.0 we need to be able to read the WSDL definition and use
that. In the meantime, I suggest we add an explicit marker to endpoint
to indicate not to expect a response.

I don't go for the none approach.... it seems counter-intuitive to me.
That seems to be telling the target what to do, not Synapse.

Paul

On 3/14/07, Asankha C. Perera <[EMAIL PROTECTED]> wrote:
Hi Gerald
> Has anyone successfully used JMSSender for one-way messages?
>
> My set-up is really simple:
> -----------------------
>        <proxy name="proxy" transports="http">
>            <target endpoint="jms_endpoint" />
>            <wsdl key="wsdl_with_one_way_op" />
>        </proxy>
> -----------------------
>
> The only special thing is that the operation defined in the WSDL and
> invoked by the client is one-way (an in-only MEP).
The problem here is that you do not define to Synapse which operation to
invoke with the WSDL, or if its IN only or IN-OUT. I guess we could
specify this within the endpoint definition, or maybe even set the
replyTo header to none to indicate that the message is not expecting a
reply?
> I experience that a one-way message (to an in-only MEP) that is sent
> to a JMS destination via JMSSender causes JMSSender to wait for a
> response message on a reply
> queue. Of course this does not make sense, and indeed the blocking JMS
> receive of that reply message ultimately times out with this error:
>
> -----------------------
> DEBUG [13 Mar 20:33:13]  JMSSender - JMSSender invoke()
> DEBUG [13 Mar 20:33:13]  JMSSender - [Client]Sending message to
> destination : TOPIC.T_INT_CLIENT_EVT
> DEBUG [13 Mar 20:33:13]  JMSSender - Waiting for a maximum of 30000ms
> for a response message to destination : QUEUE.JMS_TQ3
> WARN [13 Mar 20:33:43]  JMSSender - Did not receive a JMS response
> within 30000 ms to destination : TOPIC.T_INT_CLIENT_EVT
> -----------------------
>
> Very briefly looking at the code for
> org.apache.axis2.transport.jms.JMSSender in the trunk it indeed seems
> that JMSSender sets the flag "waitForResponse" to true regardless of
> whether the messages is one-way (to an in-only MEP) or not.
>
> Can anybody confirm my understanding of the behaviour of JMSSender?
Yes, you are correct. It is time we updated the JMS transport with the
current open issues.

asankha

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




--
Paul Fremantle
VP/Technology, WSO2 and OASIS WS-RX TC Co-chair

http://bloglines.com/blog/paulfremantle
[EMAIL PROTECTED]

"Oxygenating the Web Service Platform", www.wso2.com

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

Reply via email to