Mills@udel.edu (02/28/89)
Folks, I was not intending to advertise it widely for a few days, but Louie Mamakos found it anyway. There is a draft of what I am calling NTP Version 2 Specification on louie.udel.edu in pub/ntp/ntp.ps along with a reformatted, but otherwise unchanged, copy of the archival article pub/ntp/ntpx.ps. While both of these should print on standard PostScript fonts (Helvetica, Times-Roman and Symbol), be advised Ventura Publisher stuffs an EOT at the end of the file, which has been known to box some printer servers. Geeze, I even got it to collate backwards, which is great for my LaserWriter, but Louie grouses that its backwards for him. Pax Wigwam. The new spec, now spoken by Fuzzballs, is intended to be a superset of the old, although I did steal four bits from the old stratum field for the new key-identifier field and add an optional message-authentication code at the end. There are still a few loose ends in the local-clock area that may not have finished twitching. Also, I think an authentication-enabled bit should be included among the peer variables. Finally, be advised the authentication scheme has been handed off to the Privacy Task Force for advice and comment, so some things may twitch there, too. There is of course no key-distribution mechanism proposed specifically for NTP and no network-management mechanism to diddle the leap-bits. I handed the last issue off to the IETF net-management mafia, but I don't think they will take it seriously. A pox on their clocks. The new version cleans up a lot of issues raised on this list, including negative delays, authentication, stability, clock selection, etc. and etc., which I will summarize later. Many of the changes are already in Louie's latest version. My hope is to finish up the changes, beat on the PTF and IETF for their blessings, get the stuff running on many/most of the existing chimers and drop the new spec on the IAB for promotion to Internet Standard, lesser-known-hero grade. Pax temporum. Dave