RE: [wsrm] Preliminary minutes of WSRM TC Conf call -050603

[<magdolna.gerendai@nokia.com>]

Hi Colleen, Sunil and all,

We have extensive mail exchanges/discussions about the duplicated identification of a reliable message, but I think, this is in connection with the mixing of different tasks between different layers.

The problem space is:
- why do we want to identify a message, 
- which message do we want to identify,
- in which layer

The unique identification of a reliable SOAP message is needed to provide the guaranteed delivery, the duplicate elimination and message ordering by the reliability layer. 
The reliable SOAP message is, what must be identified, passed from the upper layer of the reliability one. SOAP messages, generated by the relaibility layer, i.e. Acknowledges, are not needed to identify with a unique identifier.

In which layer the identification should be defined ? 
1. there exists already an identification method, specified by another layer and WS-RM just reuse it
2. the WS-RM specification defines it inside the reliability layer only 
3. the WS-RM define two level identification 

1. There is no standard, specified (general) message identification, what could be re-used by all other layer specifications as a reference. I have some doubt, whether it's needed at all, because likely it's simpler if all layers define their own identification of the messages and need not be aware, what the other layers do. 

2. This solution would be the clearest. The reliability layer inserts it's own identification of the message and removes it, when passes up to the next layer.

3. If WS-RM defines a "general" or multipurpose identification of a message, (agreed with Sunil) how can we gauarantee, that anyone else will use it (exclusively). 

br,
Magdolna  


-----Original Message-----
From: ext Sunil Kunisetty [mailto:Sunil.Kunisetty@oracle.com]
Sent: May 07,2003 19:04
To: Colleen Evans
Cc: wsrm
Subject: Re: [wsrm] Preliminary minutes of WSRM TC Conf call -050603



 Hi Colleen,

Colleen Evans wrote:

> Hi Sunil,
> I see a few potential issues with the proposed opton:
>  i) Get rid of MId and have GId/Sid set. Gid could be optional & SId
>     is mandatory. If Gid is present, SId is the unique in that GId. If not,
>     SId should be globally unique.
>
> (1) RM would use a GId/SId set to uniquely identify messages, while other features /
> extensions in the same SOAP message may use a different mechanism(s).  Could get
> pretty messy and difficult to manage and synchronize.

 Is this not a general issue? Even if we use MId, what is guarantee that other extensions
 use the same tag, they may use their own Ids.

 Until there is a standard (& universally accepted) for references & MessageIDs for
 Web Services, we will have this problem. (No, WS-Addressing is not a solution :))

>
>
> (2) Even where sequencing is not required, a GId/SId set would be required to uniquely
> identify messages for other RM features.

 Can you elaborate further? I think we could overload the semantics of SId and solve
 the problems unless I'm missing something.

 Again that's just one of my random thoughts. I'll ponder more on that myself.

 -Sunil



>
>
> It seems preferable to provide a message identification mechanism that isn't coupled
> with sequencing or any particular feature.  That was the intent in the current
> WS-Reliability spec where MessageID is in the MessageHeader with other 'generic'
> information such as From, To, etc., and GroupID and SequenceNumber are in the
> MessageOrder header and only used for ordering.
>
> Colleen