Re: [LEAPSECS] UTC Redefinition Advanced
Rob Seaman said: >> The reason I haven't been involved in this thread up to now is that I spent >> the last week in a place where apparent solar time and official clock time >> were about 7800 seconds apart. It was quite curious being near the tropics >> and yet at 08:30 the sky was still dark. > > The basis of civil timekeeping is *mean* solar time. Irrelevant - mean solar time was *also* about 7800 seconds from official clock time. The point is that the residents don't seem to have a problem with being almost 10% of a day out of sync with the sun. > Apparent time is a red herring. I only mentioned apparent time because I happened to come across a sundial - the sort where you act as the gnomon. >> Yet the residents seem to cope quite happily without their world coming to >> an end. > The corollary is that individuals (such as these tropical residents) who are > unaware of the ins-and-outs of either atomic or astronomical timekeeping are > quite unlikely to discuss the subject at all. Absence of evidence is not > evidence of absence. I'm not talking about drift, but simply about the fact that people cope with large differences between MST and LLT. And I have reason to believe that, if these residents were bothered about the issue, they would have raised it as part of their negotiations with central government over the last few years. (Just to be clear, I'm talking about a western European democracy here, not a totally different system which we're unfamiliar with.) -- Clive D.W. Feather | If you lie to the compiler, Email: cl...@davros.org | it will get its revenge. Web: http://www.davros.org | - Henry Spencer Mobile: +44 7973 377646 ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
Rob ... you're absolutely right. Eliminating leap seconds WOULD introduce a secular trend ... IF (and here's the part I haven't convinced you about yet) no other means for compensating for it is implemented, FOREVER AND EVER. The Pisa architects are in a meeting: "Let us not worry about subsidence - we can always deal with that issue later." FOREVER AND EVER is how long it will take for a hundred member countries to agree on something that every single clock on earth has to be engineered to deal with. Instituting a new time standard is not an EASY project like a manned mission to Mars. It DIFFICULT project like trying to negotiate a Palestinian-Israeli resolution ten times over. -paul ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Nov 2, 2010, at 11:05 AM, Rob Seaman wrote: On Nov 2, 2010, at 7:06 AM, Clive D.W. Feather wrote: The reason I haven't been involved in this thread up to now is that I spent the last week in a place where apparent solar time and official clock time were about 7800 seconds apart. It was quite curious being near the tropics and yet at 08:30 the sky was still dark. ... Eliminating leap seconds would introduce a permanent secular trend (no matter how small). So the 30 messages in my in-box this evening arose from a belated reply to I message I posted a week ago. Oh, well. Rob ... you're absolutely right. Eliminating leap seconds WOULD introduce a secular trend ... IF (and here's the part I haven't convinced you about yet) no other means for compensating for it is implemented, FOREVER AND EVER. This is not what the ITU is discussing. They are not making policy FOREVER AND EVER. They are dealing with a much more limited question: are occasional, sporadic, and unpredictable leap seconds, that much of our infrastructure has difficulty dealing with, an appropriate means of dealing with the secular trend TODAY? Society -- and computers -- don't seem to have much problem with leap years, which occur on a preset schedule every 97 years out of 400. Why shouldn't leap seconds be constructed the same way? If the ITU makes the decision to allow DUT1 to exceed 0.9 second, then some decades from now the ITU could reintroduce leap seconds -- with years of advance notice -- on a regular once- or twice-a-year schedule. In the meantime, as long as DUT1 remains under a couple of hundred (which would take centuries to happen), the public is apt to be less inconvenienced than they are with the present system. - Jonathan ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Nov 2, 2010, at 7:06 AM, Clive D.W. Feather wrote: > The reason I haven't been involved in this thread up to now is that I spent > the last week in a place where apparent solar time and official clock time > were about 7800 seconds apart. It was quite curious being near the tropics > and yet at 08:30 the sky was still dark. The basis of civil timekeeping is *mean* solar time. Apparent time is a red herring. (What's "red herring" in other languages?) Eliminating leap seconds would introduce a permanent secular trend (no matter how small). Periodic effects or static offsets (no matter how large) are completely different phenomena. Apologies (as always) for repeating myself. > Yet the residents seem to cope quite happily without their world coming to an > end. This is the world's only forum for discussing the specific question of eliminating leap seconds. Is it surprising that participants would have strong opinions on the subject? The corollary is that individuals (such as these tropical residents) who are unaware of the ins-and-outs of either atomic or astronomical timekeeping are quite unlikely to discuss the subject at all. Absence of evidence is not evidence of absence. Presumably any who actually did think the world at risk of ending would seek a broader audience :-) Rob ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
Jonathan E. Hardis said: > Time zones with the width of an hour are generally > acceptable, which indicates that the public might accept DUT1 as large > as 1800, or so. The reason I haven't been involved in this thread up to now is that I spent the last week in a place where apparent solar time and official clock time were about 7800 seconds apart. It was quite curious being near the tropics and yet at 08:30 the sky was still dark. Yet the residents seem to cope quite happily without their world coming to an end. -- Clive D.W. Feather | If you lie to the compiler, Email: cl...@davros.org | it will get its revenge. Web: http://www.davros.org | - Henry Spencer Mobile: +44 7973 377646 ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
In message <6f0a7636-bc9c-4617-b925-ef3849744...@noao.edu>, Rob Seaman writes: >Leap seconds are a means to an end. The end won't vanish when the >leap seconds do. But didn't that end vanish a long time ago, as in "more than 100 years" ago ? Humans used the suns position to reckon time since times before history, fair enough. But only because it was the best they had. At every turn since, where a better method have been discovered, it has been applied instead. In the later half of the 1800's were mass produced clockwork became widespread, people relied on the clocks even though they were later (se below) found to have no precise correlation to earth rotation angle, but tended to be set to whatever the locals preferred. With the advent of mass transportation in the shape of trains, this was found to suck, and using the new electric telegraph, the clocks were synchronized to whatever degree possible. Since that point in time, the clocks have been synchronized to whatever the proper authority stipulates, with no regards to solar time, mean or otherwise. Dropping leap seconds amounts to the central authority adopting the same method of time-counting as everybody else uses, and has used since the dawn of history: 24 hours to a day, 60 minutes to the hour, 60 seconds to the minute. The suns position in the sky has nothing to do with this, that is in the hands of the local(-ish) politicians who legislate your timezone. So yes, leap seconds were an means to an end, and now that astronomers have had a further 40 years to catch up with the mechanization of time-reckogning, it is time to drop that hack which handicaps everybody else. Poul-Henning PS: You still have not answered my question: Why did you use UTC when you knew it was the wrong timescale for your astronomical applications ? -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 25, 2010, at 12:29 AM, Poul-Henning Kamp wrote: > In message , > Sanj > eev Gupta writes: > >> I am still opposed, in principle, to letting NTP time (for example) diverge >> from The One True Cosmic Time; > > What "One True Cosmic Time" would that be ? Both sides (assuming this discussion really splits only two ways) are speaking loosely here. See Einstein, etc, for anything resembling "cosmic time". You will come up empty-handed. > One where astronauts on Mars would have to monitor earths rotation in order > to add or remove spurious seconds to their timescale ? Astronauts on Mars (in the unlikely event this happens during our lifetimes) will likely keep time just like robots on Mars (and their handlers back on Earth) do - using Martian mean solar time. See previous threads. Removing the opportunity for intercalary adjustments won't make it easier to synchronize disparate timescales. > Or one which, given a precise enough frequency standard and knowledge about > your relativistic whereabouts you can tell what time it is, without caring > what a particular and erratic piece of rock does in another part of the > cosmos ? (Again, see Einstein who himself punted on simultaneity.) Yet again we've simply established that there are two flavors of timekeeping - intervals and Earth orientation. Adding the orientations of additional planets on the one hand and more stringent space timekeeping requirements on the other only emphasizes that the two things are different. Civil timekeeping - on Mars or Earth - is tied to the planet's mean diurnal rotation period. Other sorts of timekeeping are tied to different things. Leap seconds are a means to an end. The end won't vanish when the leap seconds do. Rob ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Mon, Oct 25, 2010 at 15:29, Poul-Henning Kamp wrote: > In message , > Sanj > eev Gupta writes: > > >I am still opposed, in principle, to letting NTP time (for example) > diverge > >from The One True Cosmic Time; > > What "One True Cosmic Time" would that be ? > My apologies, I left out the "smiley". I know there is no such "Time", I was being tongue-in-cheek. -- Sanjeev Gupta +65 98551208 http://www.linkedin.com/in/ghane ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
In message , Sanj eev Gupta writes: >I am still opposed, in principle, to letting NTP time (for example) diverge >from The One True Cosmic Time; What "One True Cosmic Time" would that be ? One where astronauts on Mars would have to monitor earths rotation in order to add or remove spurious seconds to their timescale ? Or one which, given a precise enough frequency standard and knowledge about your relativistic whereabouts you can tell what time it is, without caring what a particular and erratic piece of rock does in another part of the cosmos ? Poul-Henning -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
Snipped some stuff. This is the best definition of the problem I have seen for quite some time. I thought I was firmly in the camp of keeping leap-seconds, but put this way, it seems to me that I was staying within an error bound, and now (taking Jonathan as an example), all that is being done is to increase the error bound. And since everyone who really cares about solar/pulsar time would have to be applying DUT1 now, the cost to them is ensuring that their software (and processes) do not have variables that chuck out values of DUT1 greater than 1s.This _is_ a real cost, no doubt, but it seems less of a problem than I assumed. I am still opposed, in principle, to letting NTP time (for example) diverge from The One True Cosmic Time; but my principles are cheap, and my engineering work does not suffer under Jonathan's formulation. -- Sanjeev Gupta +65 98551208 http://www.linkedin.com/in/ghane On Mon, Oct 25, 2010 at 07:42, Jonathan E. Hardis wrote: > There is merely a reexamination of the presumption that DUT1 has to be > kept to less than 0.9 seconds, which -- when you stop to think about it -- > is a rather arbitrary requirement. > > Let's get quantitative. What magnitude of DUT1 would be tolerable to you? > > Based on your postings here, I presume that you're fine with 0.9 seconds. > One could also, arbitrarily, limit it to 0.6 seconds by having the IERS > declare leap seconds every few months, both + and -. However, since I > haven't heard you suggest that, may I presume that you would agree with me > that DUT1 need not kept that small? Can we agree that reasonable people can > discuss what an appropriate bound on DUT1 should be for the purposes of > civil time? > > What about the public at large? In most of the world, we're long past the > point of arguing that sundials are divine and time zones are the work of the > devil. Time zones with the width of an hour are generally acceptable, which > indicates that the public might accept DUT1 as large as 1800, or so. > Certainly no one today gets bent out of shape because sundial time in > Boston is 1200 seconds different than sundial time in Washington, DC, or > that sundial time in Los Angeles is 1000 seconds different than sundial time > in San Francisco. > > So, hypothetically, what would happen if we had no leap seconds for the > next 100 years? There are people who have analyzed how the deceleration of > the Earth's rotation will affect the need for leap seconds -- and I'm not > one of them. Let me make a simple guestimate that with leap seconds > occurring about once every 18 months, in the next 100 years there would need > to be about 80 of them to maintain DUT1 at 0.9 seconds or less. This means > that, without any leap seconds, sundial time at JFK Airport in New York > would become what sundial time at Liberty Airport in Newark is now, and > sundial time in Long Beach would become what sundial time in Santa Monica is > now. DUT1 up to 100 isn't likely to cause the public much if any heartburn. > > The point is that, even if we went 100 years and did nothing, the magnitude > of the effect would be so small as to be inconsequential for most if not all > purposes of civil time. However, I don't hear people talking in terms of > doing nothing in the next 100 years. In international metrology definitions > tend to change with a natural time constant (tau) of about 30 or 40 years -- > a unit of time called a "career." And 100 years is about 2 or 3 tau. The > people at the ITU today couldn't stop their successors from changing UTC > even if they wanted to. Their successors can and will have a reasoned > debate on the best way, in their time, to take account of the fact that -- > to use your expression -- the SI second is not 1/86,400 of a day. > > The debate today is not about setting a course for all future human > history. It is only about whether -- in our time -- keeping DUT1 to 0.9 > seconds is worth the grief that it causes. > > So, to repeat what I told you at the outset of this thread, if you believe > that leap seconds are the best technical approach, long term, for keeping > DUT1 below some threshold, your challenge is to make the world safe for leap > seconds. > >- Jonathan > ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
> There is merely a reexamination of the > presumption that DUT1 has to be kept > to less than 0.9 seconds, The questionaire of the US itc did not sound like this. It sounded every bit like a scrapping of leap seconds altogether. > Let's get quantitative. What magnitude of > DUT1 would be tolerable to you? If it is decided that DUT1 should not drift indefinitely then we need leap seconds to correct it since it would be impossible at this point to introduce a different system. But let's say for argument sake we make this 3.0 seconds: Then what? I.e. Can you follow through with your point? -paul Sent from my BlackBerry® by Boost Mobile -Original Message- From: "Jonathan E. Hardis" Sender: leapsecs-boun...@leapsecond.com Date: Sun, 24 Oct 2010 19:42:01 To: Leap Second Discussion List Reply-To: Leap Second Discussion List Subject: Re: [LEAPSECS] UTC Redefinition Advanced On Oct 24, 2010, at 3:29 AM, Rob Seaman wrote: > But this isn't the discussion we've been having for ten years - and > we aren't the ones to convince. On this list we have speculated > widely on possibilities of all sorts, but the entire time a > relentless and inflexible and closed-door campaign has been carried > out by nonparticipants of this list to redefine UTC without leap > seconds. So, let me parse this. You have no idea what's been discussed in "closed-door" meetings, so your only point of reference is what been bantered about on this mailing list for ten years. That becomes the reference truth. I have no idea what's been discussed in the closed-door meetings either, but at least I've discussed the subject with some of the experts on it, so I've heard what's in their heads. There is no thoughtless and nefarious scheme to decouple civil time from solar time for the remainder of civilization, nor is there a move afoot to salt the earth so that there could never be leap seconds again. There is merely a reexamination of the presumption that DUT1 has to be kept to less than 0.9 seconds, which -- when you stop to think about it -- is a rather arbitrary requirement. Let's get quantitative. What magnitude of DUT1 would be tolerable to you? Based on your postings here, I presume that you're fine with 0.9 seconds. One could also, arbitrarily, limit it to 0.6 seconds by having the IERS declare leap seconds every few months, both + and -. However, since I haven't heard you suggest that, may I presume that you would agree with me that DUT1 need not kept that small? Can we agree that reasonable people can discuss what an appropriate bound on DUT1 should be for the purposes of civil time? What about the public at large? In most of the world, we're long past the point of arguing that sundials are divine and time zones are the work of the devil. Time zones with the width of an hour are generally acceptable, which indicates that the public might accept DUT1 as large as 1800, or so. Certainly no one today gets bent out of shape because sundial time in Boston is 1200 seconds different than sundial time in Washington, DC, or that sundial time in Los Angeles is 1000 seconds different than sundial time in San Francisco. So, hypothetically, what would happen if we had no leap seconds for the next 100 years? There are people who have analyzed how the deceleration of the Earth's rotation will affect the need for leap seconds -- and I'm not one of them. Let me make a simple guestimate that with leap seconds occurring about once every 18 months, in the next 100 years there would need to be about 80 of them to maintain DUT1 at 0.9 seconds or less. This means that, without any leap seconds, sundial time at JFK Airport in New York would become what sundial time at Liberty Airport in Newark is now, and sundial time in Long Beach would become what sundial time in Santa Monica is now. DUT1 up to 100 isn't likely to cause the public much if any heartburn. The point is that, even if we went 100 years and did nothing, the magnitude of the effect would be so small as to be inconsequential for most if not all purposes of civil time. However, I don't hear people talking in terms of doing nothing in the next 100 years. In international metrology definitions tend to change with a natural time constant (tau) of about 30 or 40 years -- a unit of time called a "career." And 100 years is about 2 or 3 tau. The people at the ITU today couldn't stop their successors from changing UTC even if they wanted to. Their successors can and will have a reasoned debate on the best way, in their time, to take account of the fact that -- to use your expression -- the SI second is not 1/86,400 of a day. The debate today is not about setting a course for all future human history. It
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 24, 2010, at 3:29 AM, Rob Seaman wrote: But this isn't the discussion we've been having for ten years - and we aren't the ones to convince. On this list we have speculated widely on possibilities of all sorts, but the entire time a relentless and inflexible and closed-door campaign has been carried out by nonparticipants of this list to redefine UTC without leap seconds. So, let me parse this. You have no idea what's been discussed in "closed-door" meetings, so your only point of reference is what been bantered about on this mailing list for ten years. That becomes the reference truth. I have no idea what's been discussed in the closed-door meetings either, but at least I've discussed the subject with some of the experts on it, so I've heard what's in their heads. There is no thoughtless and nefarious scheme to decouple civil time from solar time for the remainder of civilization, nor is there a move afoot to salt the earth so that there could never be leap seconds again. There is merely a reexamination of the presumption that DUT1 has to be kept to less than 0.9 seconds, which -- when you stop to think about it -- is a rather arbitrary requirement. Let's get quantitative. What magnitude of DUT1 would be tolerable to you? Based on your postings here, I presume that you're fine with 0.9 seconds. One could also, arbitrarily, limit it to 0.6 seconds by having the IERS declare leap seconds every few months, both + and -. However, since I haven't heard you suggest that, may I presume that you would agree with me that DUT1 need not kept that small? Can we agree that reasonable people can discuss what an appropriate bound on DUT1 should be for the purposes of civil time? What about the public at large? In most of the world, we're long past the point of arguing that sundials are divine and time zones are the work of the devil. Time zones with the width of an hour are generally acceptable, which indicates that the public might accept DUT1 as large as 1800, or so. Certainly no one today gets bent out of shape because sundial time in Boston is 1200 seconds different than sundial time in Washington, DC, or that sundial time in Los Angeles is 1000 seconds different than sundial time in San Francisco. So, hypothetically, what would happen if we had no leap seconds for the next 100 years? There are people who have analyzed how the deceleration of the Earth's rotation will affect the need for leap seconds -- and I'm not one of them. Let me make a simple guestimate that with leap seconds occurring about once every 18 months, in the next 100 years there would need to be about 80 of them to maintain DUT1 at 0.9 seconds or less. This means that, without any leap seconds, sundial time at JFK Airport in New York would become what sundial time at Liberty Airport in Newark is now, and sundial time in Long Beach would become what sundial time in Santa Monica is now. DUT1 up to 100 isn't likely to cause the public much if any heartburn. The point is that, even if we went 100 years and did nothing, the magnitude of the effect would be so small as to be inconsequential for most if not all purposes of civil time. However, I don't hear people talking in terms of doing nothing in the next 100 years. In international metrology definitions tend to change with a natural time constant (tau) of about 30 or 40 years -- a unit of time called a "career." And 100 years is about 2 or 3 tau. The people at the ITU today couldn't stop their successors from changing UTC even if they wanted to. Their successors can and will have a reasoned debate on the best way, in their time, to take account of the fact that -- to use your expression -- the SI second is not 1/86,400 of a day. The debate today is not about setting a course for all future human history. It is only about whether -- in our time -- keeping DUT1 to 0.9 seconds is worth the grief that it causes. So, to repeat what I told you at the outset of this thread, if you believe that leap seconds are the best technical approach, long term, for keeping DUT1 below some threshold, your challenge is to make the world safe for leap seconds. - Jonathan ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
Jonathan E. Hardis wrote: > Once you get over the hurdle of allowing DUT1 to be larger than 0.9, many > additional possibilities open up. You could concentrate the pain by having a > "leap minute" once a century rather than a "leap second" every year or two. > Alternatively, you could declare leap seconds 50 or 100 years in advance, > which presumably would make their implementation easier and more > straightforward. But this isn't the discussion we've been having for ten years - and we aren't the ones to convince. On this list we have speculated widely on possibilities of all sorts, but the entire time a relentless and inflexible and closed-door campaign has been carried out by nonparticipants of this list to redefine UTC without leap seconds. The consensus of the only public meeting on this topic (Torino, 2003) was ignored. That consensus? Simply: call it something other than UTC in that case. Those ten years could have been much better spent. > Or, society could decide that having a leap-anything isn't worth the bother. Society can decide all sorts of things. The real world has a way of having the last say. Warner Losh wrote: > While some take that as things are OK with leap seconds, those people that > have real-time systems cringe. It occurs to me that "real-time" is a rather strange term... ...anyway, "real-timey-ness" is an orthogonal concept to the timescale used. One has to question whether any software system on Earth can be regarded as trustworthy if leap-seconds (purely a representational issue - see previous threads) are a lost cause. As science and technology professionals, should we have patience with a line of reasoning that proceeds from the premise that technology is crap? ...and should a group that fancies itself the "precision timing community" really be pinning all its hopes and dreams on subverting UTC and decommissioning TAI? Truly bizarre. Rob ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
In message: <1010232041.aa28...@ivan.harhan.org> msoko...@ivan.harhan.org (Michael Sokolov) writes: : P.S. This E-mail message has been composed and sent using 1979 : technology. I'm using 1985 technology here: emacs :) Warner ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
Warner Losh wrote: > 2010 is a radically different world than 1970 when > leap seconds were invented. Then clearly the right solution is to abolish and ban all technology invented after December 31, 1979! MS P.S. This E-mail message has been composed and sent using 1979 technology. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
One point that the leap second haters make is that leap seconds are hard. There are a number of reasons for this, but the biggest one is acceptance. Leap seconds are hard, so why implement them right, people will just reboot, or ntp will slew the time difference out. It just doesn't matter enough, we're close enough and the disruptions are rare. While some take that as things are OK with leap seconds, those people that have real-time systems cringe. Because of this "crap is ok" attitude, nobody gets leap seconds right. In fact, they get them wrong in an amusing number of different ways. This "blind eye" to the problem perpetuates the problem, even as more and more systems need to get things right. 2010 is a radically different world than 1970 when leap seconds were invented. Since there are only a vanishingly small number of systems that get them completely right, I would argue that the ITU's proposal merely recognizes the rampant 'de-facto' standard of time keeping: people only get it right in the absence of leap seconds. While this is unfortunate for the syncrhonization of the directiont he earth points to a time base that everybody uses, it is the only thing that implementors can agree on enough to interoperate with and get right. Warner ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 23, 2010, at 1:48 PM, Hal Murray wrote: How many of these systems CURRENTLY properly handle leap seconds? How many of these cell phones and space systems and digital devices "buried beneath Antarctic ice" CURRENTLY are built to a specification that a minute can contain either 59, 60, or 61 seconds? Or that when a leap second occurs, it occurs at midnight only in the UTC+0 time zone? (4:00 PM in the afternoon in California) Why is that such a big deal? At worst, they reboot the system and it gets back in sync with the loss of some data. Okay. At 4:00 PM PST on December 31 -- one hour before the close of business on the last business day of the year -- you would have folk on the West Coast reboot all of their financial systems, their communications systems, their "smart grid" electrical systems, and whatever else happens to sync to UTC. Yeah, I know that we've had a lot of leap seconds over the years, and people have dealt with them one way or another -- mostly improvisationally, not following standards. But it's becoming tougher and tougher to deal with. To reiterate, my point is that people who think leap seconds are a good idea (and that includes me, by the way), seem to be complaining at the wrong end of the value chain. Having leap seconds is counterproductive if they choke the infrastructure. Take a "time out," on declaring leap seconds, direct the frustration towards fixing the infrastructure, and then some years from now have the discussion again on whether leap seconds would be a good idea or not. I think the more important issue would be calculating differences in times that straddle a leap second. (I'm not an astronomer.) Here's the issue. There's a number called "DUT1," which is the difference (in seconds) between UTC, an atomic time scale, and UT1, a solar-based time scale. Right now, DUT1 is constrained by agreement to be 0.9 s, or less, in magnitude (+/-). The proposal is to allow DUT1 to grow larger than 0.9 s, and that would affect systems (the few that care about it, mostly astronomical) that cannot handle these larger numbers. Once you get over the hurdle of allowing DUT1 to be larger than 0.9, many additional possibilities open up. You could concentrate the pain by having a "leap minute" once a century rather than a "leap second" every year or two. Alternatively, you could declare leap seconds 50 or 100 years in advance, which presumably would make their implementation easier and more straightforward. Or, society could decide that having a leap-anything isn't worth the bother. - Jonathan ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
> How many of these systems CURRENTLY properly handle leap seconds? How > many of these cell phones and space systems and digital devices "buried > beneath Antarctic ice" CURRENTLY are built to a specification that a > minute can contain either 59, 60, or 61 seconds? Or that when a leap > second occurs, it occurs at midnight only in the UTC+0 time zone? (4:00 > PM in the afternoon in California) Why is that such a big deal? At worst, they reboot the system and it gets back in sync with the loss of some data. I think the more important issue would be calculating differences in times that straddle a leap second. (I'm not an astronomer.) -- These are my opinions, not necessarily my employer's. I hate spam. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On 22 Oct 2010, at 18:16, Rob Seaman wrote: > On Oct 22, 2010, at 9:43 AM, Matsakis, Demetrios wrote: > >> I have now heard from two sources that the revised ITU-R draft >> recommendation TF.460-6 passed a major hurdle in Geneva last week. It will >> be sent by SG7 to the January 2012 Radiocommunication Assembly meeting. > > "Passed a major hurdle"? The phrasing from the Executive Report of Working > party 7A indicates a state they describe as "deadlocked": > > "It became quite clear the issues involved were not technical issues > and the Working Party was deadlocked on non-technical issues. The path > forward to resolve the issue and come to consensus was not apparent. The > only course of action that appeared to be open was to submit the documents to > the Study Group for resolution." It sounds like that deadlock is now irrelevant since they passed the proposal to SG7 who resolved to pass it on to the general assembly. Tony. -- f.anthony.n.finchhttp://dotat.at/___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On 2010-10-23, at 02:14, Jonathan E. Hardis wrote: > > You're free to insert an inexpensive interface box between your data source > and the systems that use the data that adds or subtracts however many integer > number of seconds you wish. For example, it would be trivial to set a > private collection of NTP servers that provide a well-documented time offset > from other NTP servers that maintain UTC. I've been told that this is not as easy as it sounds. At least it would require a protocol change, or else you will end up with incompatible but (protocol-wise) indistinguishable sets of servers. Endless grief will ensue. But in my opinion the whole discussion on who would suffer and how much is quite irrelevant and something of a red herring as it distracts from the real issue: how on earth it is possible to even consider such a fundamental change in an existing time scale in mid-flight. The whole idea that it is reasonable for the ITU to ponder, let alone decide this is quite absurd. UTC was designed to stay near UT. Refining the mechanism through which UTC tracks earth rotation would be quite reasonable. Demolishing the mechanism is insane. In the back of my head I still expect to wake from this bad dream some time. > In other words, if you want a time scale that preserves leap seconds, you're > perfectly free to maintain one yourself for astronomical purposes -- whether > or not the rest of the world follows suit. Well, if you want a time scale that has no leap seconds, you're perfectly free to switch from UTC to that time scale. And if you are tied to "legal time", you should get your legislator to switch time scales (or drop your obligation). Im my world, systems that have problems with leap seconds are either repaired or switched to TAI, but now the ITU is pushing for a copout "solution" that affects the whole world. --N ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 9:05 PM, Jonathan E. Hardis wrote: > Now comes the moment of self-appraisal. This "moment" has come a dozen times before on this mailing list(s). > Leap seconds have been around since 1972 (IIRC). Leap seconds are a means to an end. Civil time is (obviously) derived from mean solar time. The ITU-R can cheat for some purposes for some period of time. However, the moon exists, tides exist, the day lengthens. See dozens of previous threads. > we have had, since 1972, an explosion of digital infrastructure that is > designed and built without regard to leap seconds. Spend even one sentence in the proposal speculating on system engineering issues related to that infrastructure. > Minutes contain 60 seconds ... period. An SI second is not 1/86,400 of a day. Pretending it is, legislating it, don't make it so. It ain't brain surgery to collect use cases, discover requirements, write them down, build trade-off matrices, perform sensitivity and risk analyses, and do all the normal system engineering that would be performed by - say - Cisco building another network switch. The ITU-R draft is an embarrassing exercise in avoiding due diligence. The ubiquity of crappy digital technology is an argument for better system engineering, not for abandoning any semblance for a Hail Mary pass. Rob ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 11:14 PM, Rob Seaman wrote: ... Clocks appear in numerous places in the workflow. It is no simple feat to coordinate all these clocks with vintages ranging over the last quarter century. GPS? Phones? Web apps? Data start at a mountaintop telescope, but flow downhill to archives and pipelines and virtual observatory portals. Redefining UTC will put a permanent kink in scientific and historical timescales, as when performing the frequent chore of combining data from different epochs - or from different telescopes. Astronomers assemble light curves from observations taken worldwide (and in space - wanna estimate the cost of a space certified interface box). These systems have to interoperate, something that is automatically provided by UTC's current definition as an approximation to mean solar time. What about space missions in progress? What about preserving a coherent data set from decade-long synoptic surveys? Telescopes aren't just the cartoon illustration of Palomar, what about radio interferometers, gravity wave detectors, neutrino telescopes buried beneath Antarctic ice? The increasingly common networks of robotic telescopes pursuing common investigations? ... Now comes the moment of self-appraisal. How many of these systems CURRENTLY properly handle leap seconds? How many of these cell phones and space systems and digital devices "buried beneath Antarctic ice" CURRENTLY are built to a specification that a minute can contain either 59, 60, or 61 seconds? Or that when a leap second occurs, it occurs at midnight only in the UTC+0 time zone? (4:00 PM in the afternoon in California) Leap seconds have been around since 1972 (IIRC). Notwithstanding those in denial on this mailing list that the ones before the COMPETES Act somehow didn't count, we have had, since 1972, an explosion of digital infrastructure that is designed and built without regard to leap seconds. Minutes contain 60 seconds ... period. If you're going to get angry, don't get angry that some at the ITU don't share your sense of esthetics. Get angry that we live in a world full of badly designed digital stuff. That's both the true problem and the solution. Work towards the day when any digital system that must be certified must meet a specification of being able to deal with leap seconds. Work towards the day when all OSes handle them gracefully, and in a uniform fashion. Work towards the day when being able to handle leap seconds is the norm, rather than the exception. At that point ... after most everything around you now has made it into the landfill ... your kids or grandkids will be in a much better position to continue the practice. - Jonathan ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 5:14 PM, Jonathan E. Hardis wrote: > On Oct 22, 2010, at 1:16 PM, Rob Seaman wrote: > >> It will, for instance, cost astronomers many millions of dollars simply to >> restore current functionality to thousands of interoperating systems. > > Oh ... come, come. > > How do these "thousands of interoperable systems" currently get the time? > > You're free to insert an inexpensive interface box between your data source > and the systems that use the data that adds or subtracts however many integer > number of seconds you wish. For example, it would be trivial to set a > private collection of NTP servers that provide a well-documented time offset > from other NTP servers that maintain UTC. > > In other words, if you want a time scale that preserves leap seconds, you're > perfectly free to maintain one yourself for astronomical purposes -- whether > or not the rest of the world follows suit. Ah! I see. We're free to inherit the crappy project management paradigm gifted to us by the ITU-R! So the fact that NTP on dozens of hosts in the observatory dome will report a different time than the laptops brought by the observers is nothing to be concerned with. That radio time signals (the "R" in ITU-R) will report a time different from our "inexpensive interface boxes" is not fretworthy. Clocks appear in numerous places in the workflow. It is no simple feat to coordinate all these clocks with vintages ranging over the last quarter century. GPS? Phones? Web apps? Data start at a mountaintop telescope, but flow downhill to archives and pipelines and virtual observatory portals. Redefining UTC will put a permanent kink in scientific and historical timescales, as when performing the frequent chore of combining data from different epochs - or from different telescopes. Astronomers assemble light curves from observations taken worldwide (and in space - wanna estimate the cost of a space certified interface box). These systems have to interoperate, something that is automatically provided by UTC's current definition as an approximation to mean solar time. What about space missions in progress? What about preserving a coherent data set from decade-long synoptic surveys? Telescopes aren't just the cartoon illustration of Palomar, what about radio interferometers, gravity wave detectors, neutrino telescopes buried beneath Antarctic ice? The increasingly common networks of robotic telescopes pursuing common investigations? Telescopes and instruments and data archives and web services are all built and maintained by different teams. This is a massive system engineering project to even begin to pretend to address all the implications. Astronomers chose to actually fix their Y2K issues, not install pivots pushing the trouble further into the future. Oh! That's what the ITU-R is poised to do. The leap seconds don't just evaporate away folks. These are areas that could and should be addressed in any coherent and even marginally complete proposal to - what? - reinvent timekeeping for the entire world. This is yet another variation on privatizing the profits and socializing the costs. Rob ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 1:16 PM, Rob Seaman wrote: It will, for instance, cost astronomers many millions of dollars simply to restore current functionality to thousands of interoperating systems. Oh ... come, come. How do these "thousands of interoperable systems" currently get the time? You're free to insert an inexpensive interface box between your data source and the systems that use the data that adds or subtracts however many integer number of seconds you wish. For example, it would be trivial to set a private collection of NTP servers that provide a well-documented time offset from other NTP servers that maintain UTC. In other words, if you want a time scale that preserves leap seconds, you're perfectly free to maintain one yourself for astronomical purposes -- whether or not the rest of the world follows suit. - Jonathan ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 3:41 PM, Poul-Henning Kamp wrote: > Second, therefore I have long time ago said that if the astronomers > thought this was a real problem, they should send a proposal saying > "Give us $CALL and we'll shut up and support you" for some Finagle > adjustment of $CALL big enough to make it worth your while. By all means, send us your contribution :-) Shouldn't such an issue be part of the proposal being debated? Why precisely is it that the one sure-fire way to provoke spirited discussion is simply to suggest that the proposal itself sucks as an example of the art of constructing a proposal? Project planning and system engineering are not mystical enterprises. Write a coherent proposal and engage in a coherent (ideally open) process. > But thirdly, I have a hard time finding more than a single astronomer > who belives this is a Big Freaking Technological Catastrophe. Ah yes, another anecdotal argument. Let's examine the only score card we do have, the "Summary of responses to the Questionnaire on a draft revision of Recommendation ITU-R TF.460.6 (Administrative Circular CACE/516)" (R07-SG07-C-0116!!MSW-E.docx, http://www.itu.int/md/meetingdoc.asp?lang=en&parent=R07-SG07-C-0116). This one page document tabulates the eight (8) "valid responses" (I don't think they included mine :-) to the questionnaire: 1) "Do you support maintaining the current arrangement of linking UT1 and UTC (to provide a celestial time reference)?" Yes 3 No 5 2) "Do you have any technical difficulty in introducing leap seconds today?" Yes 3 No 5 3) "Would you support the revision of Recommendation ITU-R TF.460-6?" Yes 5 No 3 4) "If it is agreed to eliminate leap seconds within 5 years after approval of the revision of Recommendation ITU-R TF.460-6, would that create technical difficulties for your administration?" Yes 3 No 5 The entire coherent basis of this discussion is these eight (8) "valid responses". Five for - Three against. (62%) Or is that right? Two of the yeses (plus all of the noses) apparently have no "technical difficulty in introducing leap seconds". That is - according to this poll commissioned by the proponents of the proposal - 40% of the supporters of the proposal are unruffled by the existence of leap seconds. Why then do they care? Absence of evidence is not evidence of absence. > Most of the ones I have talked to, admittedly mostly europeans, > claim that this is no big deal, since they already cannot use UTC > as Earth orientation estimator, without the fetching DUT1 and other > earth orientation data from IERS over the internet also. Astronomers are power users of both atomic time and solar time. The actual timescales that appear in various ways in diverse empirical investigations are derived from these in involved ways. The proposal will change how the details of the algorithms and distribution systems work. There is a significant cost to that, far beyond (in one man's educated opinion) the cost of Y2K to my community. I understand this aspect of the issue doesn't affect you personally. Why then are you arguing about it? And, oh yeah - apparently they have similarly covert plans to deprecate TAI as well as UTC. > For years now, we have heard your continuous thunder about how "all > of astronomy would be badly affected", but appearantly you cannot > even get these likely doomed astronomers to give a reasonable precise > estimate of the impact ? So let me understand. A mailing list is instituted to discuss precisely the issue of ceasing the issuance of leap seconds. Since this will affect systems I am responsible for (directly and indirectly), I participate in this list. You may liken email to thunder, but nothing would stop you from skipping over my messages in blissful disregard...or maybe you did, because you are misrepresenting my (indeed tediously many) arguments over the issue(s). A number of us have pursued the task of bringing this problem to the broader attention of our community. It happens to be an obscure problem, but with broad implications. It is also the type of issue that falls in the gap between the computer science side of the community and the astronomy side of the community - a niche carved out of a niche. I am confident that the message will spread more widely as the implications become clear to more members of the community. I can't control how long that will take. > Anecdotal evidence could seem to indicate, that might be because > most of them found it easier to just preemptively fix the issue, > if it even existed in the first place, than to join your sentimental > crusade for Astronomys Proper Role In Timekeeping. Wrongo. The world's astronomical software community, a few hundred individuals, will gather for our annual meeting in Boston in a few weeks. There are precious few of us who labo
Re: [LEAPSECS] UTC Redefinition Advanced
In message <0254d452-98bf-4a46-a63d-94aeb158e...@noao.edu>, Rob Seaman writes: >Ok - so you weren't criticizing my estimate of the scale of the >economic impact on my community, but rather were suggesting that >my estimate isn't formal enough? No, that is also not what I said. I called it "handwaving" and I mean "handwaving". First, I firmly belive that just the salary expenses of implementing, ignoring or otherwise dealing with leapseconds, in what we can call "the non-astronomer world" exceed your puny estimate every single damn time we have a leap-second. Second, therefore I have long time ago said that if the astronomers thought this was a real problem, they should send a proposal saying "Give us $CALL and we'll shut up and support you" for some Finagle adjustment of $CALL big enough to make it worth your while. But thirdly, I have a hard time finding more than a single astronomer who belives this is a Big Freaking Technological Catastrophe. Most of the ones I have talked to, admittedly mostly europeans, claim that this is no big deal, since they already cannot use UTC as Earth orientation estimator, without the fetching DUT1 and other earth orientation data from IERS over the internet also. For years now, we have heard your continuous thunder about how "all of astronomy would be badly affected", but appearantly you cannot even get these likely doomed astronomers to give a reasonable precise estimate of the impact ? Anecdotal evidence could seem to indicate, that might be because most of them found it easier to just preemptively fix the issue, if it even existed in the first place, than to join your sentimental crusade for Astronomys Proper Role In Timekeeping. So yes, by now I do consider your dire forecasts of economic ruin unsubstantiated handwaving... Or to put it more bluntly: You and what army ? Poul-Henning -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
Ok - so you weren't criticizing my estimate of the scale of the economic impact on my community, but rather were suggesting that my estimate isn't formal enough? How does this differ from my criticisms (repeated year after year) of the informality of the ITU-R process? The onus is on the ITU-R to carry out a thorough, diligent, professional process. They aren't even close to meeting typical standards of due diligence. Rob -- On Oct 22, 2010, at 1:33 PM, Poul-Henning Kamp wrote: > In message <56b60f42-f2dc-4ea5-88fd-2a0c2aa0a...@noao.edu>, Rob Seaman writes: >> On Oct 22, 2010, at 11:08 AM, Poul-Henning Kamp wrote: > >> "Many millions of dollars" is certainly not hyperbole. [...] > > I called it handwaving, not hyperbole. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
In message <56b60f42-f2dc-4ea5-88fd-2a0c2aa0a...@noao.edu>, Rob Seaman writes: >On Oct 22, 2010, at 11:08 AM, Poul-Henning Kamp wrote: >"Many millions of dollars" is certainly not hyperbole. [...] I called it handwaving, not hyperbole. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 11:08 AM, Poul-Henning Kamp wrote: > Rob Seaman writes: > >> It will, for instance, cost astronomers many millions of dollars simply to >> restore current functionality to thousands of interoperating systems. I >> guess that is not technical enough. > > A) Doesn't sound technical to me (Economy possibly ?) > > B) "Many millions of dollars" ? (Handwaving ?) Of course it is an off-the-cuff estimate. It is hard to get management attention on such issues years in advance. We did get one estimate on the record from a system engineer at a midsize telescope of $3M to update software systems for such a change. Sounds about right to me, but imagine this estimate is high by a factor of 10. That still amounts to $300,000 for one telescope out of dozens of similar aperture and out of hundreds of large professional astronomical facilities of all types and thousands of smaller systems...and then there are all the amateur telescopes, planetaria, data centers, space missions... "Many millions of dollars" is certainly not hyperbole. Several of us worked on the Y2K remediation for astronomical software systems - a significant effort in our community. This will dwarf Y2K for astronomy. The hierarchy of working parties and committees of the ITU-R have failed miserably at due diligence. That it is difficult to get people's attention on an obscure issue is no excuse for forsaking the responsibility. Rob ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
In message <29673354-328c-48f0-98e1-99e9062e1...@noao.edu>, Rob Seaman writes: >It will, for instance, cost astronomers many >millions of dollars simply to restore current functionality to thousands >of interoperating systems. I guess that is not technical enough. A) Doesn't sound technical to me (Economy possibly ?) B) "Many millions of dollars" ? (Handwaving ?) -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
Re: [LEAPSECS] UTC Redefinition Advanced
On Oct 22, 2010, at 9:43 AM, Matsakis, Demetrios wrote: > I have now heard from two sources that the revised ITU-R draft > recommendation TF.460-6 passed a major hurdle in Geneva last week. It > will be sent by SG7 to the January 2012 Radiocommunication Assembly > meeting. At the Radiocommunication Assembly only countries that belong > to the ITU-R can vote and a 75% majority is required for passage of a > recommendation. I don't have the wording, but I presume it calls for > the elimination of all future leap seconds after several (5?) years > notice. "Passed a major hurdle"? The phrasing from the Executive Report of Working party 7A indicates a state they describe as "deadlocked": "It became quite clear the issues involved were not technical issues and the Working Party was deadlocked on non-technical issues. The path forward to resolve the issue and come to consensus was not apparent. The only course of action that appeared to be open was to submit the documents to the Study Group for resolution." It is unremarkable that the people pushing the initiative would regard the objections as "non-technical", although why that should make them negligible is unclear. It will, for instance, cost astronomers many millions of dollars simply to restore current functionality to thousands of interoperating systems. I guess that is not technical enough. However, in what sane world view does acknowledging the existence of a deadlock correspond to clearing a hurdle? The obvious "course of action" is to drop the corrosive proposal. Rob Seaman National Optical Astronomy Observatory ___ LEAPSECS mailing list LEAPSECS@leapsecond.com http://six.pairlist.net/mailman/listinfo/leapsecs
