Re: Corrected one-way MEP proposal

I think this is basically very good.  I do have one concern on faults:

<original section="6.4.2">
Abnormal operation during a One-way message exchange is be caused by a 
failure to transfer the message. Such failure might be silent at either or 
both of the sending and receiving SOAP nodes involved, or might result in 
the generation of a SOAP or binding-specific fault (see 6.4.4 Fault 
Handling). Also, during abnormal operation the SOAP nodes involved in the 
message exchange might differ as to whether the message exchange completed 
successfully. 
</original>
<original section="6.4.4">
This MEP makes no claims about the disposition or handling of SOAP faults 
generated by the either SOAP node. 
</original>

<proposed section="6.4.2">
Abnormal operation during a One-way message exchange is be caused by a 
failure to transfer the message. Such failure might be silent at either or 
both of the sending and receiving SOAP nodes involved, or might result in 
the generation of a SOAP or binding-specific fault.  Note, however, that 
such faults are not in general made available to nodes other than the 
generating node (see 6.4.4 Fault Handling).   Accordingly, when such 
errors are encountered, the SOAP nodes involved in the message exchange 
might have differing awareness of the success or failure of the 
transmission.
</proposed>
<proposed section="6.4.4">
Faults generated during operation of this MEP are made available at the 
generating node, but are not transmitted through the network;  because 
this MEP provides for transmission of exactly one message from sender to 
receiver,  a receiving node that faults when processing a message MUST NOT 
transmit that fault to the sender of the original message. 
(Nothing prevents the use of additional SOAP interactions to transmit to 
the originator information extracted from such a fault, but any such 
transmission is beyond the scope of this MEP.  Note also that this MEP 
provides no standard mechanism by which a receiver can determine an 
address to use when trasmitting such a nonstandard fault response.)
</proposed>

The original strikes me as the sort of spec that a sympathetic reader can 
use to justify what's intended, but which doesn't come out and say what we 
mean:  if you send the fault back, you're not using this MEP.  This MEP 
provides for exactly one message in all cases. Period.  So we >do< make a 
claim about the disposition of SOAP faults:  per this MEP, you don't send 
them anywhere.  I think we need to say that.

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Wednesday, 29 March 2006 22:33:42 UTC