W3C home > Mailing lists > Public > public-ws-async-tf@w3.org > March 2005

Message ID vs. context

From: David Hull <dmh@tibco.com>
Date: Tue, 15 Mar 2005 23:43:09 -0500
To: public-ws-async-tf@w3.org
Message-id: <4237B95D.60906@tibco.com>

The discussion of message correlation I've seen so far deals with 
correlating one message directly with another, in particular a reply 
with its request.  There is another well-known approach, however, namely 
correlating messages with some "context" entity.  In the case of simple 
message correlation, the initiating message itself serves as a context 
marker, but it's also possible for an initial operation to create or 
obtain a context ID and for subsequent messages to reference it.  This 
is common practice in many real-world systems, and there are standards 
in my mammoth "to read" pile (e.g., WS-Context) that aim to address this.

The immediate question is, is there any need or desire to broaden the 
concept of Message ID to something like "Context ID" or "Correlation 
ID", or is it OK just to treat context as a separate issue?  I'm leaning 
towards the latter.  For example, a given request/reply may exist within 
the context of some larger conversation.  Basic message correlation ties 
the reply to the request, and some other ID ties the whole operation to 
the larger conversation.  But on the other hand maybe all this is saying 
is that the same message can belong to multiple contexts at once.

 From a WSA point of view, it's seems OK to stick with message 
correlation.  Establishing a context and then referencing it in two 
messages just to handle something simple like request/reply sounds like 
serious overkill.
Received on Wednesday, 16 March 2005 04:43:42 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Wednesday, 16 March 2005 04:43:42 GMT