[LEAPSECS] Changing the name of UTC
imp at bsdimp.com
Fri Oct 17 16:09:48 EDT 2014
On Oct 17, 2014, at 10:46 AM, Sanjeev Gupta <ghane0 at gmail.com> wrote:
> On Fri, Oct 17, 2014 at 9:33 PM, Warner Losh <imp at bsdimp.com> wrote:
> I’m afraid that if there are two official time scales, both maintained,
> and both propagated due to contract language, then the “new” one
> without leap seconds would be this oddball thing that nobody actually
> implements or that some people implement and others don’t and we’d
> be left with a mess. Perhaps not the same mess we have today, but
> a new, different mess.
> I agree with your problem definition, but I do not think your solution to redifine "UTC" is the best way forward.
I don’t think I’ve posited this as “my” solution. It is a solution, and it has pros and cons.
> I have seen examples of people (software programers) who coded 18:25 when they meant 18:15, to refer to the quarter hour after 6pm. The solution to this is (in my view) not redefining an hour to be 100 mins, but re stressing that units matter. And this has happened multiple times, sometimes repeatedly within the same team (English was not their native language, but that is neither here nor there)
This example doesn’t make any sense. UTC has failed to be implemented correctly time and time again. This failure
is rooted in standards bodies deciding that leap seconds were too hard and nobody should do them.
This stems from redefining the minute, btw, in 1972 by the creation of minutes that have 61 seconds. For thousands of years (ok, hundreds) prior to that, no minutes had anything but 60 seconds. And that is a very persistent notion to change.
> Unfortunately, in most cases, my pointing out the error was met by, "Oh, OK, we will fix that, stop nitpicking".
> I am not saying the current status is not fraught with corner cases, and your solution to drop Leap Seconds may well be the least evil method. It still pains me to have to say "UTC before 2021 was loosly coupled to the sun, but now it floats freely”
Dropping leap seconds is one solution. Announcing them more than 6 months in advance is another. Changing from “UTC was an observational calendar to an atomic calendar” is another way to say it, but pejorative in the other way.
But I agree with Steve Allan’s point made earlier that there’s two different types of time, and trying to conflate them isn’t working and forcing one to win isn’t going to work any better when we pick a different one to win.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
More information about the LEAPSECS