- From: <noah_mendelsohn@us.ibm.com>
- Date: Thu, 20 Jan 2005 18:37:35 -0500
- To: Harris Reynolds <hreynolds@webmethods.com>
- Cc: Harris Reynolds <hreynolds@webmethods.com>, public-ws-addressing@w3.org, "'Vinoski, Stephen'" <Steve.Vinoski@iona.com>, vikasd@yahoo.com
FWIW, my intuition is that the pressures to make EPRs URIs go up if EPR's are likely to be persistent and stored in places like UDDI directories. At the very least, the pressures to have a URI for each possible EPR (I.e. having both an XML and a URI form for each reference), seem to increase. The Web obviously gains in power when persistent references to resources of long term interest are expressed as (or at least easily inter-converible with) URIs. The need for transient cookie like things to be URIs may be somewhat lower. Just something to think about. -------------------------------------- Noah Mendelsohn IBM Corporation One Rogers Street Cambridge, MA 02142 1-617-693-4036 -------------------------------------- Harris Reynolds <hreynolds@webmethods.com> Sent by: public-ws-addressing-request@w3.org 01/20/2005 12:05 PM To: "'Vinoski, Stephen'" <Steve.Vinoski@iona.com>, Harris Reynolds <hreynolds@webmethods.com>, vikasd@yahoo.com, public-ws-addressing@w3.org cc: (bcc: Noah Mendelsohn/Cambridge/IBM) Subject: RE: EPR to UDDI Mapping Steve, Although I think ERPs will often be transitory, I agree that this is not always the case. My understanding is that when reference properties are involved (for example in the stateful interaction use cases mentioned in DavidO's proposed solution of issue #1) those reference properties could easily change for different interactions. This would also be true in a truly asynchronous call as well where the ReplyTo could change at any time. Where ERPs would be static and more resistant to change is the case where it only contains a URI that is the network address. In these cases I would agree that they wouldn't change (hopefully) and users may want to store them in some way (either using UDDI or some proprietary mechanism). In short I agree that we don't want to limit EPRs to dynamic cases, although I think that often times they will be used in that kind of environment. Harris Reynolds webMethods, Inc. -----Original Message----- From: public-ws-addressing-request@w3.org [mailto:public-ws-addressing-request@w3.org] On Behalf Of Vinoski, Stephen Sent: Wednesday, January 19, 2005 8:49 PM To: Harris Reynolds; vikasd@yahoo.com; public-ws-addressing@w3.org Subject: RE: EPR to UDDI Mapping I'm not sure why an EPR would have to be limited to being used only in transitory cases. If it's limited in that way, IMO it just means that something else quite like an EPR will need to be devised for the non-transitory cases. That seems like a mistake. --steve -----Original Message----- From: Harris Reynolds [mailto:hreynolds@webmethods.com] Sent: Wednesday, January 19, 2005 1:15 PM To: 'vikasd@yahoo.com'; public-ws-addressing@w3.org Subject: RE: EPR to UDDI Mapping There isn't currently a mapping from WSA to UDDI, although I am not sure that one will be needed. A lot of times an EPR will be part of an interaction. Node A sends B a message and includes a ReplyTo EPR specifying where a response should be sent. I am still not convinced that EPR are intended to be "stored" in the same way that services are saved to a UDDI Registry. EPRs seem to be more transitory in nature. The discussion on physical vs. logical addresses is interesting in this context. From my perspective the physical address (transport specific location) should be stored, but the logical address (the ERP) likely will not need to be. It will be interesting to see what happens in practice in this area. As a working group we will have a lot of educating to do to ensure the specification is used as designed. Harris Reynolds webMethods, Inc. -----Original Message----- From: public-ws-addressing-request@w3.org [mailto:public-ws-addressing-request@w3.org] On Behalf Of Vikas Deolaliker Sent: Wednesday, January 19, 2005 12:24 PM To: public-ws-addressing@w3.org Cc: vikasd@yahoo.com Subject: EPR to UDDI Mapping Is their a specification that show how a service can discover an existing EPR? Is their a UDDI binding defined for WSA-Core? In general, how can one discover/search for an EPR without an hallway conversation? Thanks Vikas
Received on Thursday, 20 January 2005 23:40:27 UTC