thoughts on fake IP addresses and clock specification

[Eric S. Raymond]

Mark Atwood <fallenpegasus at gmail.com>:
> I am not at all a fan of using weird unroutable IPv4 addresses to specify
> clock types. This is an example of Dr Mills being entirely too clever by 3
> halves, and being far far too embedded in the very early history of the
> internet.
> 
> It breaks layer abstraction hygiene in all sorts of ways that make me itch.

*Emphatically* agreed.  It has been on my mind for a while that this
misfeature (and the associated horrible mess around refids) must die -
namespace cleanup was in the original tech plan I wrote back around
January.  But as a vague blue-sky sort of thing, because it's going to
take a major (and incompatible) rev of the NTP protocol to get there.

That rev needs to happen anyway for IPv6, so there's an opening.  One
of my personal goals in relationship to this project since the
beginning is to have earned a seat at the table and a voice in the
discussion when serious design work on NTP protocol version 5 begins.

There's a plausible future in which NTPsec becomes the principal
testbed for v5, because few potential implementation groups are likely
to be able to match the number of FTEs that CII can afford to fund.
I want to be here for that and up to my eyeballs in it.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>