Re: [TSP] time is a monotonically increasing function ?

[Denis Pinkas <Denis.Pinkas@xxxxxxxx>]

Thomas,

> Denis, 
>
> On Wed, Dec 18, 2002 at 05:01:10PM +0100, Denis Pinkas wrote:
>
> > > Do you think it is worth explicitly stating the fact that
> > > time is a monotonically increasing function ? 
> > >
> > > I.e. TSA, when synchronizing to an external reference clock 
> > > (e.g. GPS, ACTS, trusted/secure NTP server) MUST always
> > > skew and _never_ step back its internal clock ?

> > No. If there is a drift, then the TSU can stop for a while to generate 
> > time-stamp tokens until synchronisation with UTC is reached again and then 
> > go on.
>
>
> Exactly, the TSU slows down its internal clock until it is synchronized
> (within the Accuracy limits) with the external reference, but _never_ 
> should the TSU step back.  Let's state this, or at least - in a more relaxed
> version -: the time in TSU produced tokens should never step back.

The sentence you propose above is not adequate. The document is not supposed 
to give details on implementations or such details should be placed in the 
security considerations section. If you want to further propose an addition 
on that aspect, please make a full proposal and propose a placement for it, 
but go off-line for a while on that aspect.

Denis

> Thomas