RE: draft-ietf-impp-pres-01

One note below.

Jon Peterson
NeuStar, Inc.

> -----Original Message-----
> From: Dave Crocker [mailto:[email protected]]
> Sent: Wednesday, January 08, 2003 7:08 PM
> To: [email protected]
> Cc: [email protected]
> Subject: Re: draft-ietf-impp-pres-01
> 
> 
[snip]
> 
> Absolute time is, well, absolute.  (Jon apparently missed the benefit
> of using universal time zone reference, so that all that must be sent
> is the ending time relative to UT. )
> 

If you want to retain the concept that a subscription operation with 0
duration signifies either 'fetch' or 'unsubscribe', based on the
subscription state at the recipient, you still need some concept of starting
and ending time. Simply sending the current time will not express duration
0, exactly. You could say that if the duration received in a subscription
operation effectively amount to zero or less than zero (because some time
has passed since the operation was created) that we should treat it as
unsub/fetch, that could be fair... but it is not quite as semantically
explicit, since uncertainty of propagation delay might make the meaning of a
'current time' unclear. I would prefer that we stick with durations.

> Internet time syncrhonization is an extremely mature and well-adopted
> technology.  It's readily available and provides pretty amazing
> precision.  So, with absolute time, we can achieve very high
> syncrhonization if we want.
> 

Without question. It is unfortunate that network time is not yet universally
used. RFC2779 also mentions that time synchronization is out of scope, and
specifically excludes a requirement that would have been predicated on time
synchronization.



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