Re: [Ntp] Antw: Re: Antw: Re: Calls for Adoption -- NTP Extension Field drafts -- Four separate drafts

[Miroslav Lichvar <[e-mail]>]

On Thu, Sep 12, 2019 at 04:03:19AM -0700, Harlan Stenn wrote:
> On 9/12/2019 3:15 AM, Ulrich Windl wrote:
> > I think it was the old Dave Mills consensus like: "It's only NTP if it
> > provides the same output for the same input". So if you think of some
> > improvements, they still would not be allowed (I had some discussions while
> > implementing PPS in the Linux kernel)
> 
> It's not only about instability.

Well, do you have any examples of instability between different
algorithms?

> It's about known and agreed-upon impulse behavior.

That doesn't seem so useful to me that it should discourage people
from implementing more advanced algorithms.

FWIW, in PTP there is no specification of how the clock should be
controlled. There is a problem with poor performance in long chains of
boundary clocks (one of the reasons why transparent clocks were
invented), but as I understand it, that is not specific to an
implementation or protocol.

-- 
Miroslav Lichvar