The new refclock directive is implemented and documented

Eric S. Raymond esr at
Mon Jun 27 04:21:45 UTC 2016

The new refclock directive is implemented and documented.  This has
had some large consequences.

* Driver numbers and 127.127.t.u addresses are no longer used in the
  documentation anywhere but in clockstat file examples - and I may
  change that shortly.  Some historical mentions of 127.127.t.u
  addresses remain.

* Driver types are now referred to by *name* in refclock directives.
  Table follows:

| Name      | Flags | Driver
|local      | D     | Undisciplined Local Clock
|spectracom | -     | Generic Spectracom Receivers
|truetime   | D     | TrueTime GPS/GOES Receivers
|irig       | D     | IRIG Audio Decoder
|chu        | D     | Radio CHU Audio Demodulator/Decoder
|generic    | TM    | Generic Reference Driver (Parse)
|mx4200     | D     | Magnavox MX4200 GPS Receiver
|austron    | D     | Austron 2200A/2201A GPS Receivers
|arbiter    | D     | Arbiter 1088A/B GPS Receiver
|acts       | -     | NIST/USNO/PTB Modem Time Services
|nmea       | T     | Generic NMEA GPS Receiver
|pps        | T     | PPS Clock Discipline
|hp58503a   | T     | Hewlett Packard 58503A GPS Receiver
|shm        | T     | Shared Memory Driver
|palisade   | TM    | Trimble Navigation Palisade GPS
|oncore     | -     | Motorola UT Oncore GPS
|jupiter    | D     | Rockwell Jupiter GPS
|dumbclock  | D     | Dumb Clock
|jjy        | T     | JJY Receivers
|zyfer      | -     | Zyfer GPStarplus Receiver
|neoclock   | -     | NeoClock4X DCF77 / TDF Receiver
|gpsd       | T     | GPSD NG client protocol

* There is no longer a separate fudge command.  In the documentation,
  the term "fudge bit(s)" is no longer used; instead, those are now
  referred to as "refclock options" or "driver options" and the
  noun/verb "fudge" is reserved for the two time offset options.

* The historical config examples in /etc have been dropped.  They were
  living on borrowed time anyway - Daniel's rewrite of the security options

I'm not going to say the new refclock syntax is frozen at this
point. If you think you have a better idea, put it on the table.
However, two rules:

1. The fscking magic driver-type numbers are *not* coming back!  (I've hated
   those since day one.)

2. If it's not simpler than what's there now, it won't get far.

There will be a *limited* open period for bikeshedding about the
driver names.  Please get that urge out of your systems now, because
once we ship a release with names in I'm going to consider them
immutable.  We really can't go breaking peoples' ntp.conf files
gratuitously, especially since it looks like we're going to have to
put them through one configuration-syntax flag day for security
		<a href="">Eric S. Raymond</a>

More information about the devel mailing list