4 ms·

God forbid somebody suggest a solution to a problem.

by dymk 4y ago

kstrauser 4y ago
When the problem is “this is too hard for us (but apparently not our peers)”, a valid solution is not “let’s just ignore it, even though our idea would cause an enormous pain in the neck for the rest of the world”.
- dymk 4y ago
  Why do you think it's not also too hard for their peers? Any company with an interest in time keeping spends tons of money dealing with time zones and leap seconds.
  - kstrauser 4y ago
    Because I haven’t seen the Google or Amazon proposal to ditch it. Perhaps they’ve written them and I don’t know about them, but I’m not aware of them. We do lots of things that are hard because doing them right is often more important than doing them easily. Switching from ASCII from UTF-8 was a pain, but we did it. Software upgrades are a pain. Security infrastructure is a pain. Timezones are OMG such a pain. But in all those cases, we collectively said “welp, guess we’ve gotta do it”. And what Facebook notably didn’t propose was a way to actually make this happen. Who’s going to project manage the global coordinated effort to migrate the planet to Facebook Time? That sounds like much more work than them just fixing their time handling.
    - Yujf 4y ago
      Someone has to propose it first. Just shooting them for that seems bad. I agree that we don't just need to do something because fb wants to do it, but it might actually be worth having a discussion over.
      - kstrauser 4y ago
        OK, so I do agree with that. It’s worth having a conversation about. But I don’t feel like this rose to the level of an actual proposal. It was very short to assert a claim with such wide-reaching implications as “we should start ignoring leap seconds”. As such, I don’t think it calls for an in-depth rebuttal. Consider: Proposer: “It’s time to leave Unicode in the past. It requires us to update every part of our system to deal with UTF-8 strings instead of much simpler ASCII, and we’re spending a lot of resources. Because it’s so hard and expensive, we should all use well-tested ASCII code. People who want to interoperate with our system can just rename themselves to use the Latin-1 alphabet.” Everyone else: “No.” Yes, it is hard, and people have spent a lot of, ahem, time and money to figure out how to manage this at scale. But there are real-world-tested approaches to dealing with the issue, and I firmly believe it’s better to work out and coordinate on the remaining rough edges than to throw the whole thing away to make a handful of engineers’ jobs easier.
        dymk 4y ago
        It's not everyone else. Most countries want to get rid of the leap second. "A small number of countries however, including Russia and the United Kingdom, want to keep the leap second." https://www.nature.com/articles/nature.2015.18855 https://www.nature.com/articles/nature.2015.18855
        ncmncm 4y ago
        That is sufficient argument, alone, to drop it. UK used to decide on which date to switch to and from "summer time" every year. "The admiralty will take up the question in coming weeks."
    - jefftk 4y ago
      It's not just Facebook: apparently in 2015 most countries wanted to drop leap seconds, though some wanted to keep them: Most countries, including China, the United States and many in Europe, favour scrapping the leap second and basing UTC on the continuous tick of atomic clocks. -- https://www.nature.com/articles/nature.2015.18855 https://www.nature.com/articles/nature.2015.18855
  - thrown_22 4y ago
    Those companies hire people who can code their way out of a paper bag. Source: I managed to make a transition from UTC to TAI in a broker trader in 2009 when I was fresh out university.
kergonath 4y ago
If the solution implies de-engineering society from first principles to satisfy a programmer’s desire for regularity, it’s more of a thought experiment than a solution. The world is messy, life is messy, and so is everything else. If that’s too much for poor programmers, they need to find another job.
- barnabee 4y ago
  Agreed. Honestly the only thing that article did is make me add “to irritate Facebook/Meta” to the list of reasons not to ditch leap seconds.
- kergonath 4y ago
  > If the solution implies de-engineering society from first principles I obviously meant “re-engineering”, but “de-engineering” works as well, in this case…
msbarnett 4y ago
What solution? Meta's entire proposal is "stop making the adjustment and let dealing with the accumulated error be some future person's problem". "Let's just kick the can down the road" isn't solving anything.
- dymk 4y ago
  TFA explains that it largely doesn't matter if we just ignore the leap second. It's additional complexity to our timekeeping systems that doesn't buy us much (if any) value. All the super-precise systems which would be impacted by being one second off ignore leap seconds anyways.
lelandbatey 4y ago
And god forbid we respond with "that's not a good solution." To restate, here's the problem, quoted from the original article: > "As an industry, we bump into problems whenever a leap second is introduced." Their suggested solution is: > "As engineers at Meta, we are supporting a larger community push to stop the future introduction of leap seconds and remain at the current level of 27, which we believe will be enough for the next millennium." Most folks are rightly pointing out that there are many other solutions that we could introduce, which wouldn't have the downsides of UTC drifting away from our wall-clock time. Facebook didn't even discuss those solutions though. An example possible other solution: programmers (especially programmers at Facebook) should stop using UTC and should instead use TAI (which is literally UTC but without leap seconds). Indeed, using UTC time as anything other than a wallclock time should trend towards the norm. Even though this is a clear tru-ism, seeing it adopted would be way harder (due to language inertia and habits) compared with just changing the standard (nevermind that changing the standard would break a ton of logic built around expectations like "midnight UTC falls on an integer multiple of 86400").
- ncmncm 4y ago
  None of those are, in fact, any kind of solution. Most are strictly worse than status quo. The purpose here is for things to get better, not worse.

God forbid somebody suggest a solution to a problem.

by dymk 4y ago

God forbid somebody suggest a solution to a problem.

When the problem is “this is too hard for us (but apparently not our peers)”, a valid solution is not “let’s just ignore it, even though our idea would cause an enormous pain in the neck for the rest of the world”.

dymk 4y ago
Why do you think it's not also too hard for their peers? Any company with an interest in time keeping spends tons of money dealing with time zones and leap seconds.
- kstrauser 4y ago
  Because I haven’t seen the Google or Amazon proposal to ditch it. Perhaps they’ve written them and I don’t know about them, but I’m not aware of them. We do lots of things that are hard because doing them right is often more important than doing them easily. Switching from ASCII from UTF-8 was a pain, but we did it. Software upgrades are a pain. Security infrastructure is a pain. Timezones are OMG such a pain. But in all those cases, we collectively said “welp, guess we’ve gotta do it”. And what Facebook notably didn’t propose was a way to actually make this happen. Who’s going to project manage the global coordinated effort to migrate the planet to Facebook Time? That sounds like much more work than them just fixing their time handling.
  - Yujf 4y ago
    Someone has to propose it first. Just shooting them for that seems bad. I agree that we don't just need to do something because fb wants to do it, but it might actually be worth having a discussion over.
    - kstrauser 4y ago
      OK, so I do agree with that. It’s worth having a conversation about. But I don’t feel like this rose to the level of an actual proposal. It was very short to assert a claim with such wide-reaching implications as “we should start ignoring leap seconds”. As such, I don’t think it calls for an in-depth rebuttal. Consider: Proposer: “It’s time to leave Unicode in the past. It requires us to update every part of our system to deal with UTF-8 strings instead of much simpler ASCII, and we’re spending a lot of resources. Because it’s so hard and expensive, we should all use well-tested ASCII code. People who want to interoperate with our system can just rename themselves to use the Latin-1 alphabet.” Everyone else: “No.” Yes, it is hard, and people have spent a lot of, ahem, time and money to figure out how to manage this at scale. But there are real-world-tested approaches to dealing with the issue, and I firmly believe it’s better to work out and coordinate on the remaining rough edges than to throw the whole thing away to make a handful of engineers’ jobs easier.
      - dymk 4y ago
        It's not everyone else. Most countries want to get rid of the leap second. "A small number of countries however, including Russia and the United Kingdom, want to keep the leap second." https://www.nature.com/articles/nature.2015.18855 https://www.nature.com/articles/nature.2015.18855
        ncmncm 4y ago
        That is sufficient argument, alone, to drop it. UK used to decide on which date to switch to and from "summer time" every year. "The admiralty will take up the question in coming weeks."
  - jefftk 4y ago
    It's not just Facebook: apparently in 2015 most countries wanted to drop leap seconds, though some wanted to keep them: Most countries, including China, the United States and many in Europe, favour scrapping the leap second and basing UTC on the continuous tick of atomic clocks. -- https://www.nature.com/articles/nature.2015.18855 https://www.nature.com/articles/nature.2015.18855
- thrown_22 4y ago
  Those companies hire people who can code their way out of a paper bag. Source: I managed to make a transition from UTC to TAI in a broker trader in 2009 when I was fresh out university.

kergonath 4y ago

If the solution implies de-engineering society from first principles to satisfy a programmer’s desire for regularity, it’s more of a thought experiment than a solution. The world is messy, life is messy, and so is everything else. If that’s too much for poor programmers, they need to find another job.

barnabee 4y ago
Agreed. Honestly the only thing that article did is make me add “to irritate Facebook/Meta” to the list of reasons not to ditch leap seconds.
kergonath 4y ago
> If the solution implies de-engineering society from first principles I obviously meant “re-engineering”, but “de-engineering” works as well, in this case…

msbarnett 4y ago

What solution? Meta's entire proposal is "stop making the adjustment and let dealing with the accumulated error be some future person's problem". "Let's just kick the can down the road" isn't solving anything.

dymk 4y ago
TFA explains that it largely doesn't matter if we just ignore the leap second. It's additional complexity to our timekeeping systems that doesn't buy us much (if any) value. All the super-precise systems which would be impacted by being one second off ignore leap seconds anyways.

lelandbatey 4y ago

And god forbid we respond with "that's not a good solution." To restate, here's the problem, quoted from the original article: > "As an industry, we bump into problems whenever a leap second is introduced." Their suggested solution is: > "As engineers at Meta, we are supporting a larger community push to stop the future introduction of leap seconds and remain at the current level of 27, which we believe will be enough for the next millennium." Most folks are rightly pointing out that there are many other solutions that we could introduce, which wouldn't have the downsides of UTC drifting away from our wall-clock time. Facebook didn't even discuss those solutions though. An example possible other solution: programmers (especially programmers at Facebook) should stop using UTC and should instead use TAI (which is literally UTC but without leap seconds). Indeed, using UTC time as anything other than a wallclock time should trend towards the norm. Even though this is a clear tru-ism, seeing it adopted would be way harder (due to language inertia and habits) compared with just changing the standard (nevermind that changing the standard would break a ton of logic built around expectations like "midnight UTC falls on an integer multiple of 86400").

ncmncm 4y ago
None of those are, in fact, any kind of solution. Most are strictly worse than status quo. The purpose here is for things to get better, not worse.