Re: IETF and ISO alignment on Time Stamping

[Paul Koning <pkoning@xxxxxxxxx>]

Roland,

Thanks for those examples.

I wonder if it is wise to include things like this in the same
standard.  The time stamp protocol as originally defined is based on
the rule that the TSA supplies and certifies the time stamp.

It seems to me that a service of the form "originator supplies a time, 
server verifies that it's 'close enough' and signs that conclusion" is 
an entirely different service.  I have my doubts on whether that
service has any merit, but whether or not it does, it doesn't seem
that it should be called the same thing.

One reason for my doubts: in this proposed scheme, if I have a
timestamp for T1 and you have one for T2, where T1 < T2, what does
this prove?  It is clear that it does not prove I generated my
timestamp earlier than you did.  The time relation proven by those two 
timestamps is actually looser than what you get from the original
specification: the request for T1 arrived at the server in the
interval T1 +/- K, and that for T2 during T2 +/- K, where K is the
"close enough" tolerance of that server.  Since K is necessarily
larger than the typical network latency, you don't get as tight an
ordering this way.

I think I'd rather see a spec that supports a single well-constructed
scheme, than an open-ended framework that can be used to construct all 
sorts of things.

	paul