[
https://issues.apache.org/jira/browse/AXIS2-2191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12473619
]
Deepal Jayasinghe commented on AXIS2-2191:
------------------------------------------
+1 for the proposal.
Thanks
Deepal
> Dispatch phase optimizations can prevent Sandesha from using the back-channel
> for acks
> --------------------------------------------------------------------------------------
>
> Key: AXIS2-2191
> URL: https://issues.apache.org/jira/browse/AXIS2-2191
> Project: Axis 2.0 (Axis2)
> Issue Type: Bug
> Components: kernel
> Reporter: Matt Lovett
>
> Chamikara and I have been disussing this issue on the sandesha-dev list.
> Essentially there is code in the DispatchPhase (around line 71) that sends a
> HTTP 202 back if the message is 1-way, or if it is 2-way with an async
> replyTo.
> However, there are cases when the WS-RM layer needs to write a SequenceAck
> message into the backchannel, and this code is running before we get the
> chance to try. We cannot move our logic earlier, as we need the dispatch
> phase to have resolved the MEP before we can make the decision.
> The only way forward we can see is to define a new constant, and avoid the
> code in DispatchPhase when the constant is set. When RM is enabled we will
> set the value onto the ConfigurationContext at the time that Sandesha is
> engaged. From then on the Sandesha handlers will take the responsibility -
> returning the HTTP 202 as soon as possible if we do not need the backchannel.
> Here's a proposal for the new constant name:
> // If this property is set to Boolean.TRUE then the Dispatch Phase should not
> try so send an early
> // transport-level ack back. Downstream handlers are expected to send an
> transport-level ack as
> // soon as they are sure that the transport level ack may be sent.
> org.apache.axis2.Constants.Configuration.DEFER_TRANSPORT_ACKNOWLEDGEMENT =
> "deferTransportAcks";
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
