Glen, Paul,
thanks for the comments. Do you really see soap-transport of http on one
level like jms with the complete ws-specs like ws-notification?
HTTP has the problem of being limited to point-to-point communications
and having only synchronous invocations. But does jms not having the
same problems? You can only send one message to one queue, the
infrastructure takes the rest (can take your messages and deliver them
to different endpoints or other queues). In my eyes, Synapse can take
that too with messages over http with a clone-mediator!
The other thing is that http expects an answer from the other partner, a
202 is enough. JMS underlies the same thing because you have to
completely send a message and e.g. get back the jms-message-id after a
successfull delivery to the queue. So the asynchronous thing only
belongs also to the infrastructure.
Do you agree with me? Or I am wrong?
Thanks,
Jens
Glen Daniels schrieb:
+1 to Paul's comments. There's a situation when using end-to-end
WS-Security to sign or encrypt a WS-RM message that passes through a
JMS transport hop (but also HTTP hops) along its path - in that case
you'll pass the WS-RM headers through, but typically the thing
directly on the other side of the JMS link doesn't actually do the RM
state machine itself.
--Glen
Paul Fremantle wrote:
Jens
In theory you could layer these two, but I don't think it makes sense,
as you effectively add double the work of making sure the message is
reliable. What is more commonly discussed is to use Synapse to bridge
between RM and JMS.
Paul
On Thu, Mar 20, 2008 at 9:45 AM, Jens Goldhammer
<[EMAIL PROTECTED]> wrote:
Hello,
I have a question according ws-rm! Does it also work with jms as
underlying application protocol? In my mind it would make sense to
have
reliablity information in the soap-message itself (to make switches
between the protocols), but jms also provides reliable aspects.
Thanks,
Jens
