Re: Precise time over time
In message [EMAIL PROTECTED], Markus Kuhn writes: Poul-Henning Kamp wrote on 2005-08-10 18:26 UTC: If you want a really disturbing experience, visit a modern robotical slaughterhouse, and while you are there, observe and think about what a one second difference could do the the tightly coordinated choreography of the robots. The problem with leap seconds is when the systems are in touch with each other, have synchronized clocks and know it, and then suddenly some of the systems insert an extra second, and some of them don't... You seem to imply that all aspects of the timing and synchronization of these systems are best synchronized to one single global coordinate system. Why would you want to do that? Because the food industry is required to provide trackability for food and the requirement is UTC time. I you look at your oat meal, you'll probably find that somewhere on the package there is a timestamp. (It may have only hours and minutes for dry stuff like that however). The same is true for the medical industry, only there the requirement is always one second precision. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.
Re: Precise time over time
On Thu 2005-08-11T14:40:10 +0200, Poul-Henning Kamp hath writ: Because the food industry is required to provide trackability for food and the requirement is UTC time. Not over here it isn't. I'm pretty sure the time stamps on the milk cartons are local time, and I don't think I care much whether the clock on the dairy was accurate to a minute. It seems to me that this one would be pretty simple to fix. If every factory started using TAI (or, horrors, GPS time) and a few lobbyists went to the government to have TAI declared as a legal time scale (just as the metric system was declared legal here well over a century ago, and it, in truth, the actual standard by which things operate) then there could hardly be as much objection as the dairies are having right now to the new Daylight Saving Time legislation. -- Steve Allen [EMAIL PROTECTED] WGS-84 (GPS) UCO/Lick ObservatoryNatural Sciences II, Room 165 Lat +36.99858 University of CaliforniaVoice: +1 831 459 3046 Lng -122.06014 Santa Cruz, CA 95064http://www.ucolick.org/~sla/Hgt +250 m
Re: Precise time over time
Surely the point about the slaughterhouse is the thought of the throat slasher getting a couple of seconds ahead of the brain stunner. As for the issue of whether the slaugherhouse needs syncing to an external clock, the point is that with the prevelance of ntp, it is just as easy, or easier, nowadays to synchronize all devices to a global time standard than it is to set up a local arbitrary clock and synchronize to that. Implicit in that is the assumption that somebody-cleverer-than-me is feeding me `the right answer' via NTP, and that the software I bought follows the sometimes complex but clear cut rules regarding such issues as leap seconds. A proposal that seems to me to go half way to satisfy both pro and anti leapers is the one where computers switch to using TAI internally. We could start thinking about that now and possibly start implementing it. If the leapers prevail, in future the date command will still output UTC+timezone, if the anti-leapers prevail, it can output TAI+ something (zero?!) + timezone. Peter.
Re: Precise time over time
In message [EMAIL PROTECTED], Steve Allen writes: On Thu 2005-08-11T14:40:10 +0200, Poul-Henning Kamp hath writ: Because the food industry is required to provide trackability for food and the requirement is UTC time. Not over here it isn't. I'm pretty sure the time stamps on the milk cartons are local time, and I don't think I care much whether the clock on the dairy was accurate to a minute. You don't. The food FDA does. The knowledge may not have left the dairy's computers, but the moment 8 people get sick in the same neighborhood, the FDA will be matching batch-numbers and asking for production lots, timestamps, production line coincidence etc etc. I know the USA is a bit less tighthly regulated, but over here the law is that the identities and addresses of all employees who have been involved in the production of a given unit of milk product SHALL be available within 6 hours of the question being asked. I have noticed that when ground beef in the USA is recalled, it is often in quantities of hundreds of tons, which might indicate that a quite granular approach is being taken, over here they usuall just cancel a few hundred kilos. It seems to me that this one would be pretty simple to fix. If every factory started using TAI (or, horrors, GPS time) and a few lobbyists went to the government to have TAI declared as a legal time scale (just as the metric system was declared legal here well over a century ago, and it, in truth, the actual standard by which things operate) then there could hardly be as much objection as the dairies are having right now to the new Daylight Saving Time legislation. You think making a such legislative change to time will be that easy in the country which has not yet gone metric ? :-) I hate to repeat this again and again, but any solution which is predicated on fixing the computers is not economically feasible in the short term. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.
Re: Precise time over time
Will you support a proposal that keeps leap-second (or -minutes), but mandates that they be determined 40 or 50 years in advance ? Determined to what accuracy?
Re: Precise time over time
On Aug 11, 2005, at 12:46 AM, Poul-Henning Kamp wrote: As I understood the situation last week, nobody in the gang here had problems with leap seconds if we got a longer warning (40-50 years). So what prevents us from writing up our own proposal to ITU ? I'm pretty sure that we could get a rather impressive list of signatures from both camps if we did a bit of lobby work in our respective communities. A reasonable idea. Of course, there is no reason to suppose that any such proposal would even get a reading from the committee. They appear to be fixated on leap hours for reasons of their own. There would be some real value, however, in determining if there actually is enough common ground to stand on. No one, of course, would object to a 50 year look-ahead in the leap second scheduling algorithm. I would prefer we cast the problem in that fashion, not as a warning. It may well be that alerts and announcements of various sorts would be issued that are responsive to the scheduling algorithm, but the underlying issue is the scheduling of leap seconds or other clock synchronization activities. Note that even under the ITU proposal various activities would continue for assuring consistency between the various time scales. The question is whether a deterministic 50 year solution exists. Would love some feedback from the Earth rotation experts as to the current and projected state of the art for making such predictions. Presumably there is some trade-off between the precision of a prediction and its latency into the future. It is likely that some horizon exists beyond which no practical level of precision is predictable. This horizon is certainly greater than six months. Is it greater than 50 years? I doubt it. The LOD (Length of Day) plots in http://www.ucolick.org/~sla/leapsecs/dutc.html suggest trends that could be characterized over a few years - maybe a decade at the outside. Note that leap seconds are a result of integrating under the LOD curve - this smoothes out the annual signal. Other periodic signals are undoubtedly also well characterized. The question of this scheduling horizon is independent of the leap second scheduling algorithm itself. I made a proposal similar to what Kamp suggests back in April 2001: http://iraf.noao.edu/~seaman/ leap. Obviously the document would have to be reworked into a form suitable for an international standard from its current expression of the opinion of a blowhard enthusiast, but the central notion is simply to use the current standard fully. Permitting leap seconds to be scheduled monthly would allow improving the tracking between UTC and UT1. (It would also force non-conforming projects to fix their systems to actually agree with the standard or perhaps to use a more appropriate time scale. I consider this a feature of my proposal. Sweeping leap seconds under the rug is not going to solve the underlying problems.) There is no reason that my proposal could not be combined with the 50 year predictive horizon to satisfy both of us. As far as the maximum permitted size for DUT1 - some think 0.9 seconds is too small. There appears to be a consensus among quite divergent thinkers here that 0.9 hours is much too big. I imagine most astronomers would be willing to entertain intermediate values. Personally I think such a discussion would be unwise without the participation of folks within the ITU leap second firewall. Any motion on the part of astronomers to unilaterally relax the 0.9s limit would simply be taken for capitulation to the ITU proposal. Without any transparency into the process, one might even think that that was precisely the point of the leap hour proposal: scare folks with an outrageous suggestion into writing a counter-proposal more to the liking of the ITU. It would be trivial, however, to convert the current UTC standard based on leap seconds into a standard based on leap minutes. My proposal suggests a scheduling algorithm that would keep DUT1 within plus-or-minus 30s in that case. It is likely that the state of geophysical understanding would not support a 50 year horizon for leap-minute predictions, but it might support 10 or even 20 years. However - are leap minutes likely to be more palatable to Kamp's caricature of a moronic Posix programmer? Will neglectful and naive project management really benefit from 20 year advance notice of a leap minute? I still argue that the way to encourage compliance to a standard is to make non-compliance a non-option. A project facing the prospect of dealing with a potential leap second monthly is a project that will actually expend the effort to get this right. A project facing an event 50 years - or even 10 years - in advance will simply ignore that event. And a culture facing a one-hour discontinuity 500 years in the future is a culture that is facing a massive Enron-style accounting disaster. Creative accounting will always get you in trouble. UTC is not broken. A fix
Re: Precise time over time
On Thu 2005/08/11 12:37:52 MST, Rob Seaman wrote in a message to: LEAPSECS@ROM.USNO.NAVY.MIL As far as the maximum permitted size for DUT1 - some think 0.9 seconds is too small. There appears to be a consensus among quite divergent thinkers here that 0.9 hours is much too big. I imagine most astronomers would be willing to entertain intermediate values. But, thinking particularly of the piece by Jan Meeus (on this list recently via Christian Steyaert), I expect that many astronomers would be quite irritated by that. However, allowing DUT1 to grow potentially to tens of seconds would be more patatable than the current leap hour proposal, so a predefined long-term extrapolation, as described by Steve Allen (Aug/5), seems possible as a compromise. However, the question that naturally arises is the required timescale of the extrapolation. A figure of 50 years seems first to have been suggested by Poul-Henning Kamp (Aug/04, My personal opinion is that 50 years seems right, 20 years might be livable) and since seems to have become set. However, I question the need for such a long extrapolation. What systems being constructed now will need to know the time to the nearest second for 50 years without the possibility of being updated? Cast your minds back to 1955; the state of technology then; what was built then that is still running now. If the extrapolation could be reduced the potential excursion of DUT1 would also be reduced. I think the idea would be much more saleable with an initial timescale of, say, 20 years, extended by 5 years every 5 years. So at any time the extrapolation would range between 15 and 20 years. Mark Calabretta ATNF
Re: Precise time over time
On Aug 11, 2005, at 7:27 AM, Peter Bunclark wrote: Surely the point about the slaughterhouse is the thought of the throat slasher getting a couple of seconds ahead of the brain stunner. That's what I love about this example - it just gets worse and worse the more you try to clarify. I see, little Sally, that you don't fully understand what I'm talking about. You see, it is R2D2's job to whack the steer over the head with a brain-stunner, and then C3P0 slashes its throat to let all the blood pour out into the open trench that flows throughout the abattoir. Watch your step now... Just think how - messy - it would be if the throat slasher went into action before the brain stunner! Blood would splatter just everywhere! And the brain stunner might then miss entirely due to the animals thrashing about in their death throws, which would make the subsequent guts eviscerator [presumably manned by Pee Wee Herman's robot from Star Tours] a real horror show. And as you surely now see, little Sally, all of these robotical slaughterhouse shenanigans are like totally the fault of those dastardly leap seconds... Rob Seaman National Optical Astronomy Observatory
