Re: [ws-rx] [NEW ISSUE]: Suggest the restricted use of anonymousURI

[Paul Fremantle <paul@wso2.com>]

Doug

I'd like to start a more concrete discussion on the use of anon-reply-to 
for two way communications.

The question is what level of reliability can be achieved in this 
scenario. All the response messages must be piggybacked onto synchronous 
transport responses. The reliability fails if there are no longer 
synchronous request messages on which to piggyback responses. However, 
this is identical to the situation where one end goes down. Before we 
deleted them, the delivery assurances talked of either messages being 
delivered reliably *or a fault being raised on one or both ends* (or 
words to that effect). In the case where the respondingRMS cannot send 
ackRequests it can raise a fault - just as it would if it could not 
complete an ackRequest against a real URI.

Although I don't think we've fully worked through the scenarios around 
this case, Stefan has promised to produce one for the interop. If that 
fails then I'd like to re-address this concern. I certainly think that 
the spec needs tightening up - especially around the Offer - to make 
this work, but I do not agree with the current proposal, which limits 
this case unnecessarily.

Paul


Doug Davis wrote:
>
> Title: suggest the restricted use of anonymous URI
>
> Description/justification:
> When an AS uses an RMS to reliably send messages to an RMD, the RMS 
> will need to be able to resend the un-acked messages at will.  If the 
> AS uses a target URL or wsa:To value such that the RMS can not, at its 
> own discretion, initiate the (re)sending of messages then the RMS 
> would be severely limited in its ability to complete its job.  To this 
> end the RM spec should discourage the use of wsa:To values that would 
> put the RMS in this situation, like the anonymous URI.  Of course, 
> there may be times that using the anonymous URI _and_ RM can work and 
> so we shouldn't totally ban the use of anonymous URI but we should 
> make people aware that w/o some other mechanism, a generic WSA+RM soap 
> stack would not be able to support this.  Note, that while this is 
> phrased in the context of wsa:To, for replies the RMD becomes the RMS 
> and the wsa:ReplyTo becomes the wsa:To - so it would mean that we'd, 
> implicitly, be discouraging the use of the anonymous URI in 
> wsa:ReplyTo when people would want responses sent reliably.
>
> Target: core
>
> Type: design
>
> Proposal:
>
> After line 441 of [1] add:
>
> Messages sent using this protocol MUST NOT use a wsa:To value that 
> would prohibit the RM Source from retransmitting unacknowledged 
> messages.  For example, using the anonymous URI, without any 
> additional transmission mechanism, would restrict an RM Source's 
> ability to re-establishing a new connection to the RM Destination when 
> a re-transmission of a message is needed.
>
> thanks,
> -Doug
>
> [1] 
> http://www.oasis-open.org/apps/org/workgroup/ws-rx/download.php/16095/wsrm-1.1-spec-wd-08.pdf 
>

-- 

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

http://bloglines.com/blog/paulfremantle
paul@wso2.com

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