ws-rx message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: Re: [ws-rx] Presence of an wsa:relatesTo part within an acknowledgementmessage
- From: Doug Davis <dug@us.ibm.com>
- To: chamikara@wso2.com
- Date: Wed, 8 Feb 2006 15:17:42 -0500
The RM spec is silent on it because
I don't believe its related (no pun intended :-). A separate
Ack message will probably not have a wsa:relatesTo header because its not
related to any other message - but that doesn't mean that it can't be -
it just means that RM does not require it to. Some other spec may
add that header for its own use but that's should have no impact on RM
at all.
thanks,
-Doug
Chamikara Jayalath <chamikara@wso2.com>
02/08/2006 12:50 PM
Please respond to
chamikara |
|
To
| Doug Davis/Raleigh/IBM@IBMUS
|
cc
| ws-rx@lists.oasis-open.org
|
Subject
| Re: [ws-rx] Presence of an
wsa:relatesTo part within an acknowledgement message |
|
Hi,
RelatesTo does not always mean that the second message is an reply,
right ? (depending on the relationshipType).
Currently the spec does not specifically say that the ack message will
not have an relatesTo value. Could some implementation choose to mention
that. (this may cause implementations that does dispatching based on the
relatesTo header to function incorrectly) (I'm talking abt acks that are
sent as seperate messages, without being piggybacked).
Thanks,
Chamikara
Doug Davis wrote:
>
> Acks are not replies and therefore the presence, or absence, of a
> wsa:relatesTo in a message
> is not influences by it at all. When an AckRequested is sent
there is
> no guarantee that an Ack will
> be sent back on any kind of response flow - it just asks the RMD to
> send an Ack - it doesn't say
> anything about when or where that ack goes. The Ack will be
sent to
> the AcksTo EPR which
> may be totally different from the replyTo EPR that carried the
> AckRequested. In cases where
> the Ack is not piggy-backed on another message, the Ack message is
> just a new one-way message.
> Does that help?
> thanks,
> -Doug
>
>
>
> *Chamikara Jayalath <chamikara@wso2.com>*
>
> 02/08/2006 12:09 PM
> Please respond to
> chamikara
>
>
>
> To
> ws-rx@lists.oasis-open.org
> cc
>
> Subject
> [ws-rx]
Presence of an wsa:relatesTo part within an acknowledgement
> message
>
>
>
>
>
>
>
>
>
> Hi All,
>
> I have a question regarding the relationship between an sequence
> acknowledgement message and the corresponding request message. I think
I
> missed a part of the discussion that happened through this list. Pardon
> me if this issue was discussed previously.
>
> The problem is weather an wsa:RelatesTo tag should be present in an
> acknowledgement message. The presence of an wsa:RelatesTo part will
make
> implementation issues such as dispatching easy. I.e. an web service
> framework will be able to identify the relationship between an ack
> message and the original application message that asked for this and
it
> will be able to easily relate them (without going to the RM level).
>
> But to my knowledge,having a wsa:relatesTo value without the
> wsa:relationshipType attribute imply that the second message is an
> 'reply'. But an acknowledment is not an application level reply to
the
> original message and it might be better to have an different value
for
> the wsa:relationshipType attribute.
>
> On the other hand somebody could argue that the wsa:relatesTo header
> should not be present in an ack message since it is an RM control
> message and since it may have no direct connection with the original
> application message (even though it trigured the RM destination to
send
> this ack).
>
> What should be the convention ?
>
> Thank you,
> Chamikara
>
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]