Re: Predictable Internet Time

Joe Touch <touch@xxxxxxx> · Tue, 3 Jan 2017 16:23:03 -0800



    On 1/3/2017 4:13 PM, Tony Finch wrote:

    
      Joe Touch <touch@xxxxxxx> wrote:

      
      >

      
      > You're right; I got it backwards. Except it admits it's
        an approximation:

      
      > "A value that approximates the number of seconds that
        have elapsed since

      
      > the Epoch."

      
      Thank you :-)

      
      This kind of confusion is why leap seconds don't work in
        practice.
    
    
    No, it's just why Unix cannot compute the time since epoch without
    the table of leap seconds.

    
    Nobody said that keeping accurate time wouldn't require external
    information.

    
      > However, when you are computing epoch time from UTC,
        you'd have to

      
      > account for the leap seconds since epoch to correct for
        the difference

      
      > in actual elapsed seconds.

      
        POSIX and NTP say you aren't allowed to
          account for leap seconds when computing seconds since the
          epoch from UTC.
      
    
    Then they're both wrong. What do you want me to say?

    
    We've understood the concept of local frames of time since the 1920s
    (or should have). All we need to do is get our specs in line with
    reality. Note that GPS does it right - it indicates time since an
    epoch and has to be corrected to convert to UTC, and includes the
    delta to UTC in its announcements.

    
    Joe