OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

ws-rx message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: RE: [ws-rx] PR33 - Re: [ws-rx] NEW Issue back-channel not defined



Paul,
Taking a quick look at the Axis 2.0 docs version 1.1... and I am
glancing at it pretty quickly...
It seems to include a pretty fat shim that wraps up a bunch of specs
into a fairly large abstraction.

It looks to me that SMTP is a bit of a misnomer in this case and is only
one of the moving parts.

Looking at the configuration documentation, it seems to implement
receiving via a pop3 client and sending through a generic email/smtp
server.

Since all mail flows through a single user account, then the "email
level" addressing characteristics seem to be fixed.  A reply will, in
this case, be returned to the sender's (which means Axis') email
account.
I am assuming that addressing headers in the SOAP message(s) will be
used for dispatch.

Since the W3C Note SOAP 1.2 Email binding was written before
WS-Addressing, and based on what Axis advertises as functionality, I
find it hard to believe that it is using that binding.  I am guessing
that the WS-A message correlation is used, but I am not sure without
reading code.

Maybe what you really are writing is an Axis binding :-)

In any case, it looks to me like a round trip req-resp involves two or
more instances of smtp and one or two instances of pop3 from a transport
point of view.  The aggregate collection of loosely associated parts
flying in formation can mimic a synchronous req-resp even though each of
the transport entities are inherently async.
The issues related to where or if firewalls exist are substantially
different from the http makeconnection use case.  Different enough IMO
to warrant that both fish not be caught in the same net.

Cheers
-bob

-----Original Message-----
From: Paul Fremantle [mailto:paul@wso2.com] 
Sent: Thursday, December 14, 2006 2:12 PM
To: Bob Freund-Hitachi
Cc: Richard Salz; ws-rx@lists.oasis-open.org
Subject: Re: [ws-rx] PR33 - Re: [ws-rx] NEW Issue back-channel not
defined

Ok now I'm really confused!!!

So Axis2 has some code that kicks in if the request comes over SMTP and 
there is either no WSA, or an anon WSA ReplyTo. All it does is behave 
like a mail client's reply button does. I did read the specs at the 
time. It seemed pretty clear the logic to follow.

Paul

Bob Freund-Hitachi wrote:
> Paul,
> Maybe I am just confused.
> It seems to me like you are using wsa:anon to be an alias for a valid
> mailto: scheme uri.
>
> Are you talking about the rfc2822 replyTo or the rfc2821 (smtp)
reverse
> path?
>
> Is a non addressable email client one that has neither a FQDN (rfc2881
> 4.1.1.1) nor an address literal (rfc2821 4.1.3)?
>
> Or are you mapping wsa:anon to a real address at the originating side
> and placing that string in the rfc2822 replyTo field?
> Or are you assuming that the rfc2821 reverse path will always work to
> send a message to the originating endpoint when the rfc2822 replyTo is
> set to wsa:anon (or missing for that matter)? 
>
> Are you assuming that a "wsa:anon" client would poll for responses via
> smtp?
> Unless rfc821's TURN command was used, the transmission direction
would
> not reverse.  I have assumed rfc2821 where TURN has been deprecated.
> Why would not the poll be something like imap or pop?
>
> Thanks
> -bob
>
> -----Original Message-----
> From: Paul Fremantle [mailto:paul@wso2.com] 
> Sent: Thursday, December 14, 2006 8:25 AM
> To: Bob Freund-Hitachi
> Cc: Richard Salz; ws-rx@lists.oasis-open.org
> Subject: Re: [ws-rx] PR33 - Re: [ws-rx] NEW Issue back-channel not
> defined
>
> Yes exactly. From my perspective a backchannel is any way I can get a 
> SOAP response back to the originator when the wsa:replyTo is
anonymous.
> Obviously this relies on the transport.
>
> In fact my definition is very simple. If you can get a response back
to 
> an anon client, then there is a backchannel.
>
> In the SMTP case the SMTP Reply-To header allows that.
>
> Paul
>
> Bob Freund-Hitachi wrote:
>   
>> What is it?
>> How does rfc2821 return a mime body on the same connection?
>> Are you thinking that a correlated response using the rfc2822
>>     
> message-ID
>   
>> sent to the replyTo address is a backchannel?
>> Thanks
>> -bob
>>
>> -----Original Message-----
>> From: Paul Fremantle [mailto:paul@wso2.com] 
>> Sent: Thursday, December 14, 2006 7:49 AM
>> To: Bob Freund-Hitachi
>> Cc: Richard Salz; ws-rx@lists.oasis-open.org
>> Subject: Re: [ws-rx] PR33 - Re: [ws-rx] NEW Issue back-channel not
>> defined
>>
>> Bob
>>
>> That is exactly why I don't agree with your definition.
>> I believe that SMTP has a backchannel.
>>
>> Paul
>>
>> Bob Freund-Hitachi wrote:
>>   
>>     
>>> No, in the case of an rfc2822 message carried over an rfc2821
>>>     
>>>       
>> transport
>>   
>>     
>>> there is no backchannel (as defined in the chris/bob joint
>>>       
> definition)
>   
>>> since rfc2821 deprecated the rfc281 TURN command.
>>> In rfc2821 there is no way that a response may be transmitted over
>>>       
> the
>   
>>> same connection as the request.
>>> -bob
>>>
>>> -----Original Message-----
>>> From: Richard Salz [mailto:rsalz@us.ibm.com] 
>>> Sent: Wednesday, December 13, 2006 8:20 PM
>>> To: Paul Fremantle
>>> Cc: ws-rx@lists.oasis-open.org
>>> Subject: Re: [ws-rx] PR33 - Re: [ws-rx] NEW Issue back-channel not
>>> defined
>>>
>>>   
>>>     
>>>       
>>>> I still don't agree that this is right. I think there may be cases
>>>>     
>>>>       
>>>>         
>>> where 
>>>
>>>   
>>>     
>>>       
>>>> there is a new transport level connection. The main point is that
>>>>         
> the
>   
>>>>       
>>>>         
>>   
>>     
>>>> response channel is transport-defined not WS-A defined.
>>>>     
>>>>       
>>>>         
>>> Hm.  For a SOAP-over-SMTP binding, would you expect the backchannel
>>>       
> to
>   
>>> be 
>>> a response message, the equivalent of the recipient invoking the
>>>     
>>>       
>> 'reply'
>>   
>>     
>>> function on its mail user-agent?  (I think the question is
>>>     
>>>       
>> interesting; 
>>   
>>     
>>> either there is no back-channel or there is only the back-channel.)
>>>       
> I
>   
>>>     
>>>       
>>   
>>     
>>> think it's up to the particular transport binding to say, tho.
>>>
>>>         /r$
>>>
>>> --
>>> STSM
>>> Senior Security Architect
>>> DataPower SOA Appliances
>>>
>>>   
>>>     
>>>       
>>   
>>     
>
>   

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

http://bloglines.com/blog/paulfremantle
paul@wso2.com
(646) 290 8050

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



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]