Daylight saving time
-
It'd be really great if there were some way to broadcast a change to the current local timezone across a subnetwork, but I've never heard of any way to make an IP phone do that.
If the IP phones in question are capable of picking up tz info from DHCP per rfc4833, you could do it by rebooting network switches (probably worth deferring that reboot until none of the connected phones was in a call).
-
You wouldn't even have to reboot the switches, you could go with a very short lease time on the phone vlan (5, 10 minutes).
Alternately, I'm pretty sure that you could use central provisioning on the phones to update the TZ, but you might generate reboots for the phones when they pick up the provisioning change (for instance, Cisco SPA 50x and 3xx phones will reboot for pretty much any provisioning change, whereas the SPA 525 generally only "refreshes" its SIP module and doesn't do a full phone reboot). The degree to which that's acceptable would depend on how many phones have computers daisy chained behind them.
-
You wouldn't even have to reboot the switches, you could go with a very short lease time on the phone vlan (5, 10 minutes).
But then you'd have to deal with switching leases during calls. Would that work?
-
No, by DHCP protocol standards, the lease should still be valid and just be reaffirmed at the halfway point on the lease. As long as the machines stay on, then no other machines can take their lease.
Depending upon how many devices you have, you could end up with a ton of DHCP traffic.
-
As referenced, the standard DHCP protocol will attempt to renew the same lease at 50% (and 75% and 87.5%) of the original lease time.
Certain older MSoft clients (Win XP, at least originally) would pitch a fit if the original lease wasn't divisible by 8 to get to a whole number of either seconds or minutes. But most modern equipment will handle things just fine -- and potentially generate a fair amount of DHCP traffic.
But if your network falls over by adding 1000 DHCP renewals every 5 minutes, I'd put forward that you probably have an underlying network issue somewhere. Plus, if I understand @lightsoff's current situation correctly, he's already generating a fair amount of NTP traffic that he could then throttle back once he doesn't have to use the NTP kludge to keep the time in sync.
-
@Intercourse said:
No, by DHCP protocol standards, the lease should still be valid and just be reaffirmed at the halfway point on the lease. As long as the machines stay on, then no other machines can take their lease.
Cool. I genuinely didn't know that. My stuff is mostly at a much higher level of networking (I rarely need to venture below the application layer). It's a good day when I learn something…
-
In technology, I find that I learn something new every day. It is a field that suits those well who like to learn.
I have said before: "Working in technology is like trying to drink from a firehose. You are one person trying to learn everything you can while thousands upon thousands of other people are out there innovating and coming up with new shit every single day. There is no possible way that anyone could know everything, it pays to be part of a good team."
-
@Intercourse said:
In technology, I find that I learn something new every day. It is a field that suits those well who like to learn.
Yes, but I've a splitting headache today which has sort of hindered a focus on learning. :-( Still, I wrote a presentation, had a meeting, deployed a new version of a server stack, and got a couple of hours in on porting some tricky code. Slow day.
I don't count finding out some trivial detail of a library/framework “learning” as they're usually so focused on the situation. Protocols though, they're much more reusable knowledge.
-
I know it doesn't fit with the character of this site, but thanks very much!
I'd really never thought of using DHCP to set timezones, my RFC search-foo is clearly crap!
Certainly worth investigating, I wonder if this is generally supported by embedded systems with no filesystem.Should be similar traffic than the existing short NTP, maybe less.
-
@Intercourse said:
No, by DHCP protocol standards, the lease should still be valid and just be reaffirmed at the halfway point on the lease. As long as the machines stay on, then no other machines can take their lease.
Most DHCP servers will also try to be consistent about handing out the same IP address, if possible, to a client whose MAC address they've seen before. If the DHCP address pool is bigger than the number of devices that come and go on the network, this lets clients get stable IP addresses even if they go offline for long enough that their leases expire.
Obviously that's not behavior that should ever be relied on, but it does make DHCP work better than you'd expect in many cases.
-
I'd really never thought of using DHCP to set timezones
You know you can use it to set the address of the preferred NTP peer as well, yes?
-
I'll have the SI second officially renamed as the "sec",
If you are doing this I propose that you also set 18:00 to be sunset, so that a summer day lasts longer, with short nights. For those near the poles you'd have to make a compromise that forces a minimal length of true seconds for night/day so that you can get a free hours sleep/work!
Though I'd make the informal unit "sec" so that the ejaculation "wait a sec!" isn't referring to a precise length of time.
-
If you are doing this I propose that you also set 18:00 to be sunset, so that a summer day lasts longer, with short nights.
Thought about that, but rejected it on the grounds of needing far too much violence done to the duration of the second. Just forcing sunrise to be 06:00 all year round everywhere only needs about as much stretch and squish as is already customarily tolerated in non-NTP-synchronized clocks. It doesn't cause the kind of wild inconsistency between the durations of daytime seconds and nighttime seconds, or the durations of winter working hours vs. summer working hours, that would be required to make sunset occur consistently at 18:00 clock time as well.
People are also already accustomed to sunset happening at later clock times in summer, including happening at very late clock times during DST.
The proposed alignment of 06:00 to sunrise confers the advantage of DST I've most often seen expressed by those who like it (more daylight available after working hours) while getting rid of the jetlag-inducing 1-hour time step that causes such horrible pain for me every year. I'm as implacably opposed to present-day DST as anybody you're ever likely to meet, but I would not object at all to every sunrise always happening at 06:00.
RESOLVED WONTFIX
I'd make the informal unit "sec" so that the ejaculation "wait a sec!" isn't referring to a precise length of time.
Thought about that option and rejected it as well, on the grounds that minutes and seconds refer to 1/60th and 1/3600th divisions of degrees as well as hours. Traditional timekeeping is based on dividing days, not compounding seconds.
There's a definite need for a precise and consistent SI unit of duration, and the second has been co-opted for that purpose. Personally I see that as a fundamental error, both on aesthetic grounds (having the fundamental SI unit be a derived traditional unit is ugly) and on technical grounds (leap seconds now need to be a thing). The commercial and personal world got on just fine with wonky Earth-rotation-based timing before atomic clocks got invented, and I'd like to see the good parts of that make a comeback.
Let scientists have secs and give the rest of us sloppy seconds.
RESOLVED WONTFIX
-
The proposed alignment of 06:00 to sunrise confers the advantage of DST I've most often seen expressed by those who like it (more daylight available after working hours)
If that was your goal, IMO you failed miserably because you made things significantly worse during winter.Compared to even now, you moved at least my sunrise significantly earlier, which moves our already depressingly-early sunset even earlier. Under your plan, sundown would happen on the summer solstice around 3pm (instead of the 4:30 that happens now, and the 5:30 that would happen with year-round DST).
If you want to fix one of the ends of the day, fix the end of it: sundown happens at 8pm or something.
And I'm in the continental US, not some exotic locale. (Edit: For a quantitative measure, you moved sunset earlier, meaning less daylight available after work, for about 7 months of the year in Chicago.)
-
Compared to even now, you moved at least my sunrise significantly earlier, which moves our already depressingly-early sunset even earlier. Under your plan, sundown would happen on the summer solstice around 3pm (instead of the 4:30 that happens now, and the 5:30 that would happen with year-round DST).
Chicago summer solstice
Sunrise: 05:15 CDT = 06:00 ALT (Alt Local Time)
Sunset: 20:29 CDT = 21:14 ALTChicago winter solstice
Sunrise: 07:15 CST = 06:00 ALT
Sunset: 16:23 CST = 15:07 ALTSunset at or after 17:00 CST this year: Jan 28 .. Nov 1 (end of CDT, CST resumes Nov 2) = 278 days; total daylight hours after 17:00 = 625; between 20-Mar and 22-Sep, 536
Sunset at or after 17:00 ALT this year: Feb 24 .. Oct 17 (daylight >= 11 hours) = 236 days; total daylight hours after 17:00 = 594; between 20-Mar and 22-Sep, 566
So adoption of ALT would mean Chicago gets days when sunset happens after the end of the working day 65% of the year instead of 76%, and total number of daylight hours after work over the whole year would reduce by 5%. On the other hand, total daylight hours after work during the warmer half of the year between spring and autumn equinoxes would increase by 5% and at no time would you ever need to get up before sparrow's fart. Calling that significantly worse is a bit of a stretch.
If you want to fix one of the ends of the day, fix the end of it: sundown happens at 8pm or something.
Fixing sunset at 20:00 instead of sunrise at 06:00 would mean that the 9 hour day length in the middle of winter would put sunrise at 11; you'd be required at the office two hours before dawn. I'd call that significantly worse.
CLOSED INVALID
-
Fixing sunset at 20:00 instead of sunrise at 06:00 would mean that the 9 hour day length in the middle of winter would put sunrise at 11; you'd be required at the office two hours before dawn. I'd call that significantly worse.
I wish my winter days were that long. I think they're only about 7 hours through December and early January…
-
There's an old joke I remember about winters in Finland: If you sleep too late, you'll not only miss sunrise, you'll also miss sunset.
-
Fuck seasonal affective disorder! Hibernate.
-
-
Fuck sunlamps! Get badly burned genitals.
-
Fuck badly burned...ok, nevermind that idea. Next time stay back from the lamp a bit.
-
Fuck badly burned...ok, nevermind that idea. Next time stay back from the lamp a bit.
Paging @algorythmics?
-
being badly burned would never be enough to stop me. even if I ended up circumcised by the time I was done, I would not pass up a sexytimes opportunity
-
Not quite. The datum for pre-tz times was traditionally solar noon at 12:00:00. My Way New Local Time Paradigm fixes 06:00:00 at sunrise
Yes, very quite. Pre-tz time that already used modernish hours was solar noon at 12, but earlier still the day indeed started by dawn.
If that was your goal, IMO you failed miserably because you made things significantly worse during winter.
When the photoperiod is shorter than waking hours than normal waking hours you end up doing similar amount of lighting no matter how it is synchronized to clock. The purpose of summer time is to minimize overlapping sleeping time with daylight and synchronizing at dawn does that well as it would allow most people wake up around dawn.
Synchronizing to sunset would do that too, but slightly worse as time when people go to sleep varies more than time when they get up. But more importantly people are built to wake up at dawn, their biological cycle is synchronized to dawn and for millennia they woke up to rosters crowing at daybreak. So synchronizing dawn to 6:00 would minimize problems people have with waking up by allowing most of them to wake up at dawn throughout the year.
Russia appears to be TRWTF, even worse than Spain. The great red peril still lives!
Couple of years back Russia decided to ditch switching time (sensible) and fix the time zones used during summer (yeah, that).
-
So synchronizing dawn to 6:00 would minimize problems people have with waking up by allowing most of them to wake up at dawn throughout the year.
I suspect you'd need a significant societal change for that to be true.. I am pretty sure you'll find that most people wake up well after 6am, and I sincerely doubt that's because sunrise is occurring after 6am1. Personally, I'd very much like to see the typical work day, store hours, event times, etc. moved earlier an hour if 6am were sunrise (and even that wouldn't be nearly enough for my tastes actually, but I'm ready to admit fault there); I think there's a lot more... "pressure" to stay up late (like I'm doing now) then there is to get up early, and most of that would remain with the fixed 6am sunrise -- meaning that there are a lot of people like me who all that would do is cause us to waste precious winter sunlight sleeping.
1 It's harder than I'd have expected to find data on this. http://www.statcan.gc.ca/pub/11-008-x/2008001/article/10553-eng.htm says Canadians have an average wakeup on workdays of 6:54am. http://www.timeuseinstitute.org/Twenty-four Hours 7June08.pdf says the median wake-up time is about 6:30am. http://www.perthnow.com.au/news/western-australia/research-shows-average-west-australian-wakes-up-at-645am/story-fnhocxo3-1227019570715 says average wake-up times range from 6:45 to 7:54 depending on country and region (the 7:54 is in France). http://snoozester.com/The-Wake-Up-Time-Report.snooze is a company that delivers wakeup calls, and says that their most common time is 6am, but that the average time is a bit after 7; but I suspect their data is probably very unrepresentative. In any case, almost all of those numbers are noticeably after 6am, and some are substantially after.
-
I did just pick 06:00 somewhat arbitrarily, on the grounds that this is the time that the sun rises on the equator when noon is fixed at 12:00.
However, most of the non-tropical world does seem to prefer some form of Daylight Saving Time. And in a DST regime, solar noon spends half the year at roughly 13:00 rather than roughly 12:00. Getting rid of the DST time step while leaving things roughly where they already are should therefore actually involve working from the idea of noon averaging to 12:30, not 12:00.
So I'd be willing to explore the possibility of adjusting ALT to fix sunrise at 06:30 rather than 06:00. Running the Chicago numbers again, that would give you
Chicago summer solstice
Sunrise: 05:15 CDT = 06:30 ALT (Alt Local Time)
Sunset: 20:29 CDT = 21:44 ALTChicago winter solstice
Sunrise: 07:15 CST = 06:30 ALT
Sunset: 16:23 CST = 15:37 ALTSunset at or after 17:00 CST this year: Jan 28 .. Nov 1 (end of CDT,
CST resumes Nov 2) = 278 days (76% of year); total daylight hours after 17:00 = 625; between 20-Mar and 22-Sep, 536Sunset at or after 17:00 ALT (daylight >= 10.5 hours) this year: Feb 13 .. Oct 28 = 258 days (71% of year); total daylight hours after 17:00 = 718; between 20-Mar and 22-Sep, 658
So although there's still a decrease in days where the sun is still up at 17:00, that's now only by 5% of the year, and is more than compensated for by substantial increases in the total number of post-worktime sunlight hours both year-round and in the warmer half.
How does that sit with you?
-
I am pretty sure you'll find that most people wake up well after 6am
05:30
It's nice when the birds protest at your alarm clock in the morning...
-
I set my phone to 5:45 and the alarm clock I've already shown on here to 6:00. Waking up often takes a long time for me.
But I have been waking up around 30-60 minutes early during the last two or so weeks.
-
I think there's a lot more... "pressure" to stay up late (like I'm doing now) then there is to get up early,
Totally. I'm amazed at how late people stay up.
Our school district has jumped on the bandwagon for moving high school start times to be later. It's going to cost the district about $5M annually due mainly to extra buses required. My belief is that the kids having problem with getting to bed at the current times will stay up a little later since they can sleep in a little later.
-
-
My belief is that the kids having problem with getting to bed at the current times will stay up a little later since they can sleep in a little later.
Well, maybe. Teen brain development naturally causes some degree of delayed sleep phase; I'd expect to see better academic performance once the later start time has had a chance to get bedded in.
-
Well, maybe. Teen brain development naturally causes some degree of <a href="http://en.wikipedia.org/wiki/Delayed_sleep_phase_disorder">delayed sleep phase;</a> I'd expect to see better academic performance one the later start time has had a chance to get bedded in.
That's the typical excuse given. I just don't buy it for the simple explanation that some people just refuse to go to bed because they might be missing something. I don't doubt this is a legit issue for some people, but I will be really shocked if we see the magnitude of benefits supporters have claimed.
-
Clearly pointless doing any experiment ever. Just ask boomzilla, he'll know what will happen.
-
Clearly pointless doing any experiment ever. Just ask boomzilla, he'll know what will happen.
Exactly.
Seriously, though, it's not like this is any sort of a cheap experiment. The district has been doing stuff like delaying teacher raises and generally complaining about how they don't have enough money. IIRC, there are some additional up front costs and $5M is the extra amount annually.
-
-
my alarm is officially 0615, but the cat has other ideas. she'd been waking me up with the tried and true method of a fang to the earlobe at 0530 for the past month and demanding i give her many scratches.
and yes i did check for fleas/ticks/empty food dishes/etc. it's not any of that.
-
she'd been waking me up with the tried and true method of a fang to the earlobe
So how many of those piercings have you used for earrings?
-
my alarm is officially 0615, but the cat has other ideas. she'd been waking me up with the tried and true method of a fang to the earlobe at 0530 for the past month and demanding i give her many scratches.
and yes i did check for fleas/ticks/empty food dishes/etc. it's not any of that.
Of course it isn't. She's a cat, therefore she is the most important thing in the world (to her), and you must tend to her every desire on demand.
-
So how many of those piercings have you used for earrings?
She hasn't actually managed to complete the piercing yet. i always wake up befroe then.
-
Of course it isn't. She's a cat, therefore she is the most important thing in the world (to her), and you must tend to her every desire on demand.
QFT
<substance
-
I'm unemployed. I never wake up.
-
-
No way man. I spend two solid days thinking: "man, come Monday, I really gotta look at ramping-up the job search..."
This is contrasted with weekdays, which are the days I spent not ramping-up the job search.
-
-
also tried. all that got me was a nice puncture mark on the hand. and more demands for scratches.
-
05:30
It's nice when the birds protest at your alarm clock in the morning...
Yup.
Of course, I'm almost always up before the alarm. You quickly learn that when the 17 year old dog wants to go out, you get up. This happens anywhere between 2a and 5a. This morning was 5:25a - YEA!!!
-
When I was in middle school I had a flock of birds outside my window that would raise a racket five minutes before my alarm went off every morning without fail. Looks like fifteen minutes before my alarm, someone let their cat out, and the cat took ten minutes to make his way to his favorite tree, which he jumped into from the railing on the balcony outside my window, scaring all the birds that roosted there overnight.
-
That is an impressive Rube Goldberg alarm...
-
... except for the math error:
Let ta = time of alarm.
Let tb = time at which birds raise racket.
Let tc = time at which cat is let out.Given:
tc = ta - 10
tb = ta - 5tc + 10 ≠ tb
-
Flag for math pedantry?
good catch though!
