[LEAPSECS] Leap seconds ain't broken, but most implementations are broken

[Steffen Nurpmeso]

Martin Burnicki <martin.burnicki at burnicki.net> wrote:
 |Steffen Nurpmeso wrote:
 |> Martin Burnicki <martin.burnicki at burnicki.net> wrote:
 |>|Ask Bjørn Hansen wrote:
 |>|> [1] As much as I dislike the leap seconds; smearing is only appropriate
 |>|> if you specifically have chosen it so they␦re not supposed to be in the
 |>|> pool.
 |>|
 |>|Agreed. Smearing in the way it is currently done by some Google servers
 |>|(and ntpd, if configured accordingly) is just a hack to workaround
 |>|problems with applications that are unable to account for leap seconds
 |>|correctly, and should only be used in closed environments.
 |> 
 |> So if you do it only in closed environments over that full control
 |> is available then why should you have your clock off by multiple
 |> hours and not slew a single second, for example?
 |
 |With "closed environments" I didn't mean that accurate time doesn't
 |matter and thus no time reference needs to be available.

Oh, that is a misunderstanding.

 |I rather thought of let's say "closed time distribution", e.g. a company
 |which has its own time server(s) rather than using public servers, so
 |the admin can decide if their server(s) should smear a leap second, and
 |all the PCs and other nodes in the company network can synchronize to a
 |smearing or non-smearing server.

So i have understood it.  For many in. e.g., Germany you of course
have the problem that the secretaries computer is very likely
turned off almost the entire span of the proposed smear, so that
the OS upon startup will likely see a very large time gap.  And
in this spirit, if you have a known set of installed applications
and they all are known to accept slewing the second, this is
likely good enough, or even better, then.

 |> It surely would have been much easier if CLOCK_TAI and the UTC
 |> offset would have been widely distributed and advocated for many
 |> years, and if the most portable standards would offer interfaces
 |> for them.  Since the problem as such doesn't seem to leave us,
 |> that is to say.
 |
 |Agreed.

Yes, and of course distribution of leap seconds as such, so to be
able to actually interpret why your clock is off by an amount that
would indicate a failure otherwise.

--steffen