Re: [Phys-L] Leap Second Computer glitches

[John Denker <jsd@av8n.com>]

On 07/02/2012 11:33 AM, chuck britton wrote:
> Many in the IT community want to do away with these inconvenient
> beasties - but I think a better plan is to learn to deal with them.

I'm not sure what "beasties" that refers to.
 -- If we are talking about gross failures of scheduling systems, 
  the only way to deal with the failures is to make them go away.
 -- If we are talking about leap seconds in connection with civil
  time, they are not going away any time soon.

That may look like a dilemma, but it's not.

There is a straightforward way to avoid any dilemma:  Do not use
civil time for timestamping, sequencing, or interval calculations.
Readily-available alternatives include TAI and TDB:
  http://en.wikipedia.org/wiki/International_Atomic_Time
  http://en.wikipedia.org/wiki/Barycentric_Dynamical_Time

I note for example that the "qmail" program uses TAI for timestamping.
This makes sense.

Sooner or later it will be considered programming malpractice to use 
UTC for timestamping, sequencing, or interval calculations.

Alas I'm not holding my breath.  There is a certain /major/ software
vendor that still sets the hardware clock to *local* civil time ... 
not even UTC ... even on laptops that can easily move from one timezone
to another.  I wonder whether they even know how to spell TAI.