On Fri, Jan 20, 2012 at 10:13 AM, Tim Bray <tbray@xxxxxxxxxxxxxx> wrote: > One consequence of your proposal, if adopted, is that there will need > to be a specification of the canonical Internet-time-to-Sidereal-time > function, so that in the long run, the time that your computer says it > is will correspond with what you observe looking out the window. The > Internet will be around long enough that this will indeed become a > problem. > > I'd want to look at that specification before getting passionate pro > or contra in this argument. -T The people who really care about this (i.e., astronomers) already use TAI to do it. I have written such code (more than once), the first thing you do is to find UT1 - TAI, then proceed with various rotations from there. 50 years ago, using UTC as an approximation to UT1 when you didn't happen to have a multi-million dollar, multi-ton, mainframe in your pocket made sense. Today, when that same power (or, actually, more) is in your smart phone (if not your toaster), it doesn't. Regards Marshall > > On Fri, Jan 20, 2012 at 6:20 AM, Phillip Hallam-Baker <hallam@xxxxxxxxx> wrote: >> If we are ever going to get a handle on Internet time we need to get rid of >> the arbitrary correction factors introduced by leap seconds. >> >> The problems caused by leap seconds are that they make it impossible for two >> machines to know if they are referring to the same point in future time and >> quite often introduce errors in the present. >> >> 1) No machine can determine the number of seconds between two arbitrary UTC >> dates in the future since there may be a leap second announced. >> >> 2) If Machine A is attempting to synchronize with machine B on a future >> point in time, they cannot do so unless they know that they have the same >> view of leap seconds. If a leap second is announced and only one makes the >> correction, an error is introduced. >> >> 3) In practice computer systems rarely apply leap seconds at the correct >> time in any case. There is thus a jitter introduced around the introduction >> of leap seconds as different machines get an NTP fix at different points in >> time. >> >> 4) Even though it is possible to represent leap seconds correctly in >> standard formats, doing so is almost certain to exercise code paths that >> should be avoided. >> >> >> Since the ITU does not look like sorting this out, I suggest we do so in the >> IETF. There is no functional reason that Internet protocols should need leap >> seconds. >> >> I suggest that the IETF plan to move to Internet Time in 2015, immediately >> after the next ITU meeting. Internet time would be TAI plus the number of >> leap seconds that have accumulated up to the next ITU decision point. So if >> UTC drops leap seconds at the next meeting the two series will be in sync, >> otherwise there will be a divergence. >> >> >> >> -- >> Website: http://hallambaker.com/ >> >> >> _______________________________________________ >> Ietf mailing list >> Ietf@xxxxxxxx >> https://www.ietf.org/mailman/listinfo/ietf >> > _______________________________________________ > Ietf mailing list > Ietf@xxxxxxxx > https://www.ietf.org/mailman/listinfo/ietf _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf