Mailing-List: contact cxf-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: cxf-dev@incubator.apache.org
Received-SPF: neutral (herse.apache.org: local policy)
Message-ID: <7b774c950703161209y49914769jbbcd13cbda949b53@mail.gmail.com>
Date: Fri, 16 Mar 2007 15:09:50 -0400
From: "Dan Diephouse" <dan@envoisolutions.com>
To: cxf-dev@incubator.apache.org
Subject: Re: Why is OutgoingChainSetupInterceptor in inbound chain ?
In-Reply-To: <FA1787F64A095C4090E76EBAF8B183E071FCAD@emea-ems1.IONAGLOBAL.com>
MIME-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_Part_79037_5749466.1174072190768"
References: 
 <E91104197DA2654DA833B6EB53B6C488025F6C46@emea-ems1.IONAGLOBAL.com>
	 <FA1787F64A095C4090E76EBAF8B183E071FCAD@emea-ems1.IONAGLOBAL.com>

------=_Part_79037_5749466.1174072190768
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

What about:
C) The outMessage/outgoing chain gets created once we know we have an output
message?

Otherwise I would prefer B.

- Dan

On 3/15/07, Glynn, Eoghan <eoghan.glynn@iona.com> wrote:
>
>
>
> Hi Ajay,
>
> I also found it confusing having the OutgoingChainSetupInterceptor
> running in the PRE_LOGICAL phase.
>
> If the response has to be sent before this phase is entered (e.g. as a
> result of a JAX-WS protocol handler killing the dispatch), then the
> relevant interceptor has to take responsibility for setting up the
> outgoing chain anyway.
>
> The confusion is only deepened by the AbstractServerFactoryBean adding
> this interceptor to the normal *in* chain, but also the fault *out*
> chain.
>
> So it would seem more logical to either:
>
> A) set up the out-chain before even the first interceptor in the
> in-chain is traversed, e.g. in the ChainInitiationObserver.onMessage()
>
> B) leave to the POST_INVOKE phase as you suggest so that there is no
> confusion, i.e. so every in-interceptor knows it must take explicit
> responsibility for setting up the out-chain if it needs to send a
> 'premature' response.
>
> Contrast with the current scenario whereby the in-interceptor would only
> need to do the out-chain setup if it runs in a phase before PRE_LOGICAL,
> or even more error-prone if its in PRE_LOGICAL phase itself but doesn't
> call addAfter(OutgoingChainSetupInterceptor.class).
>
> Cheers,
> Eoghan
>
>
> > -----Original Message-----
> > From: Paibir, Ajay [mailto:ajay.paibir@iona.com]
> > Sent: 15 March 2007 15:35
> > To: cxf-dev@incubator.apache.org
> > Subject: RE: Why is OutgoingChainSetupInterceptor in inbound chain ?
> >
> >
> > Dan
> >
> > If a inbound interceptor wants to change the flow from
> > inbound to outbound or operate on outbound message/chain then
> > it has to take explicit responsibility of setting up the
> > outbound message and chain.
> > Why do it before hand when it is not known if the outbound
> > message is normal repsonse or fault?
> > Also it is puzzling that out chain/message were created in
> > PRE_LOGICAL not as early as Protocol Phases or ChainObserver.
> >
> > The use case was to include only interceptor operating b/w
> > Phase.PRE_LOGICAL  to Phase.INVOKE in inbound and on outbound
> > include interceptor from Phase.SETUP to Phase.POST_LOGICAL.
> > And my custom interceptor will setup the outbound chain in
> > POST_INVOKE phase. But what I observed was the outgoing chain
> > was already evaluated in the inbound side.
> >
> > Ajay
> >
> > -----Original Message-----
> > From: Dan Diephouse [mailto:dan@envoisolutions.com]
> > Sent: 15 March 2007 14:57
> > To: cxf-dev@incubator.apache.org
> > Subject: Re: Why is OutgoingChainSetupInterceptor in inbound chain ?
> >
> > I think the point of puttoing the
> > outgoingchainsetupinterceptor early in the inbound chain is
> > to create an out message and outgoing chain asap.
> > This allows further interceptors to operate on that
> > message/chain if they wanted to.
> >
> > Whats your use case for changing it?
> >
> > - Dan
> >
> > On 3/15/07, Paibir, Ajay <ajay.paibir@iona.com> wrote:
> > >
> > > Hi,
> > >
> > > I noticed the OutgoingChainSetupInterceptor is part of the inbound
> > > interceptor chain operating in the PRE_LOGICAL phase. It does the
> > > following things
> > > a) Sets up the outbound message before even the message is
> > dispatched
> > > to the implementor.
> > > b) Copy Message properties from inbound to outbound message
> > > c) Sets up the outbound interceptor chain.
> > > d) Sets up outbound fault message.
> > >
> > > I think a) and c) are done in a eager manner and should be deferred
> > > further down the inbound chain or if possible to start of outbound
> > > chain.
> > >
> > > So i am proposing to move (a),(b)  to
> > ServiceInvokeInterceptor and (c)
> >
> > > to OutgoingChainInterceptor which intiates the outbound
> > chain. Leave
> > > the
> > > (d)  to the fault observer namely OutFaultChainInitiatorObserver.
> > >
> > > Thus eliminating OutgoingChainSetupInterceptor.
> > >
> > > Any suggestions, comments? I volunteer to do the changes if
> > it makes
> > > sense.
> > >
> > > Regards
> > > Ajay
> > >
> > >
> >
> >
> > --
> > Dan Diephouse
> > Envoi Solutions
> > http://envoisolutions.com | http://netzooid.com/blog
> >
>


-- 
Dan Diephouse
Envoi Solutions
http://envoisolutions.com | http://netzooid.com/blog

------=_Part_79037_5749466.1174072190768--