Ajith, I don't think you're deeply mislead, it seems there's quite a lot of
us who can't understand what's wrong with having a Synapse specific message
which delegates some calls to an Axis2 MessageContext.
interface SynapseMessage {
String getTo();
...
}
SynapseMessageImpl {
MessageContext mc;
public SynapseMessage(MessageContext mc) {this.mc = mc}
String getTo() { return mc.geTo(); }
...
}
This way the SynapseMessage interface can include methods that are on the
Axis2 MC and also have additional methods specific to Synapse. I don't see
this as reinventing Axis2 or wasteful inefficient doubling up of method
calls,looks more like standard OO design to me.
...ant
Ajith Ranabahu <[EMAIL PROTECTED]> on 13/11/2005 05:51:59
Please respond to [email protected]
To: [email protected]
cc:
Subject: Re: [Axis2] SOAPMessageContext
Ok, I don't want to look like the mad scientist who comes up with the
craziest theory always but I humbly think I'm talking about the same issue
of allowing the SOAPMsgCtx to co-exist without corrupting (or rather
effecting) the Axis2 system. However I also feel that I lack knowledge of
some discussions which were in the mailing lists (Silly me) so I won't push
the msgCtxtFactory theory anymore.
BTW is this SOAPMC needed to be implemented as part of Axis2 ? The original
mail seems to imply that the SOAP specific methods were specifically for
Synapse, in which case the particular 'Ctxt' needs to be part of Synapse
and not Axis2!
Am I deeply mislead or missing a crucial piece of logic here ?
On 11/13/05, Sanjiva Weerawarana <[EMAIL PROTECTED] > wrote:
Hi Paul,
> If I am in Synapse I am looking at a message on the wire. And I have
> to decide which way it is going. So in general I have a complex
> calculation to make:
>
> 1. Look at the transport (i.e. is it an HTTP response)
> 2. See if it has a relatesTo header
> 3. Look at the To: header (maybe it is anon)
> etc.
>
> If it is an In Out* mep, then the In.isResponse=false and the
> out.isResponse is true for both following messages.
Let's do this- when we figure out how Axis2 and Synapse will work
together (at the message level, not architecturally) let's see whether
there's a way to have such a flag.
I think the important point is that from Synapse's point of view, we
want to know whether the message is going to the service vs. coming from
the service; we don't particularly care whether its a "response" or not.
That is, for an outbound MEP, I assume we'll want to apply the rules for
messages going out of the service and not the ones for messages coming
in, right? In which case talking about "is it a response or not" is the
wrong topic.
Sanjiva.
--
Ajith Ranabahu
