[LEAPSECS] Crunching Bulletin B numbers (POSIX time)

[Ian Batten]

On 21 Feb 11, at 1117, Tony Finch wrote:


> On Mon, 21 Feb 2011, Ian Batten wrote:

> 

>>> Ascii timestamps do not have a way to include leap second information.

>> 

>> But they trivially could.

> 

> Is there any standard way to do so? If there is I would love a citation!

> 

> Time zone information is recorded in timestamps as an offset


Except when it isn't.  RFC822 and its successors have had a good influence, and mailsystems usually get this right.  But ctime(3) doesn't, and outside mail systems it's rare (see my previous message) to find systems that record the timezone symbolically, never mind numerically.  



> , so if the

> software is set to an unexpected timezone or has stale timezone tables,

> you can still understand what time it thought it meant. The only way to

> get a similar lack of ambiguity wrt leap seconds is to include the TAI-UTC

> offset in the timestamp.


There's no reason why you couldn't extend numeric tz offsites to include leapseconds, as in "Mon, 21 Feb 2011 11:17:16 +000021".  


> 

> But note I am assuming that you want to represent an absolute point in

> time (say, an astronomical event) rather than an event identified by the

> position of the hands on a clock.


I suspect that most people want the latter.  The interests of those that want the former need to be accommodated, but I suspect that the communities that need solar time, were that to become unhinged from civil time, are smaller in number than those that need civil time whatever it may represent.


>  In every-day affairs it

> is very unusual to schedule future events for absolute points in time, but

> instead they are scheduled for a particular local time in a given place

> regardless of timezone or leapsecond disruption.


A property Posix time doesn't have. 

ian