RE: comments on draft-ietf-impp-cpim-pidf-05

At 10:02 PM 8/27/02 -0400, Peterson, Jon wrote:
>Yes, I agree, it is a transplant from other XML protocols, and unfortunately
>only a partial one. To reiterate a point that I perhaps made too obliquely
>in my initial note, the mustUnderstand mechanism in SOAP relies on the
>generation of SOAP faults which report back the circumstances of the failure
>to the sender, so that the sender can tailor their messages as appropriate.
>This allows a negotiation down to a lowest common denominator. The absence
>of such feedback is the problem with mustUnderstand as a negotiation
>mechanism in PIDF.

[My 3rd and hopefully last response on this topic...]

Yes, this is true.

But consider that mustUnderstand here is a feature of the presence 
notification *format*, which is defined separately from the notification 
protocol (abstraction).  Should the lack of one mean that we don't try to 
plan for evolvability of the other?

Also, some of the specific CPIM-compatible protocol designs may (do) have 
fault reporting mechanisms, so mustUnderstand allows for functional 
enhancement within a protocol environment, with fallback triggered by fault 
reports from a CPIM gateway.  Later versions of CPIM may extend this 
capability.

I favour keeping mustUnderstand because it is a cheap mechanism that lays a 
key bit of groundwork for evolving the CPIM protocol framework, allowing us 
to concentrate our early efforts on a simple specification.

#g


-------------------
Graham Klyne
<[email protected]>




  [reminder: [email protected] for non-technical discussions, please]