Thanks Tammo for the detail explanation. -- Thilini
On Sat, Nov 16, 2013 at 1:05 AM, Tammo van Lessen <[email protected]>wrote: > Hi Thilini, > > I think my concern was against the idea to use the WS-A RelatesTo header > with BPEL correlation. IMO these are orthogonal concepts. If no exotic > relationshipTypes are defined a WS-A RelatesTo header identifies the > message ID of the request, making the message itself a reply to the request > message. This implies to me that both messages implement a request/response > MEP. Thus I believe that these headers should be added and interpreted by > the Web service stack or the integration layer but should not be passed to > the BPEL correlation mechanisms. The WS-A WSDL binding for instance would > provide the means the enable an asynchronous transport via HTTP using WS-A > headers to do the correlation between request and response completely on > the SOAP/transport level. Headers are usually meant to contain control > information for the messaging stack (e.g. addressing, security), where as > the body (which contains the payload) is targeted to the application, in > this case the BPEL engine. Making the headers available to the BPEL > language is not defined in the standard and an extension of ODE. I believe > it should only be used if there is no other way deal with the headers. > > The BPEL correlation, in contrast, is meant to provide means for a message > correlation using business values. So instead of artificial message IDs, > correlation relies on meaningful values like the orderID, caseID, or any > other set of values that are part of the body. In addition to that, such a > correlation identifier is not restricted to one request-response-pair but > can span a complex conversation between partners. This makes it a very > powerful and flexible tool. > > I hope this makes it a bit more clear. > > Cheers, > Tammo > > > > > On Thu, Nov 14, 2013 at 7:38 AM, Thilini Ishaka <[email protected] > >wrote: > > > Hi Tammo, > > > > According to the previous discussion on 'using WS-A headers in bpel > > correlation' at [1], it says, that's not a good idea conceptually. Could > > you please elaborate more on that. > > > > [1] > > > > > http://mail-archives.apache.org/mod_mbox/ode-user/201112.mbox/%[email protected]%3E > > > > I checked with ODE 1.3.5 and this functionality did not work. I'm yet to > > check that with ODE 1.3.6 > > > > Thanks in advance > > Thilini > > > > > > -- > Tammo van Lessen - http://www.taval.de >
