[LEAPSECS] Crunching Bulletin B numbers (POSIX time)

[Paul Sheer]

I think that it is extremely important that this issue is
understood. So let's try another example....


Ian Batten wherefore sayeth,

>  Yes, I can see leap-second would cause a problem there, except (a) the

>  engineering design is insane and (b) even if it isn't, TAI is ready

>  and waiting to solve their needs.  Has anyone, anywhere, written a

>  system so stupid and not been locked up?



Only machines that GENERATE timestamps NEED to have accurate system
time. On all other machines clocks are free to drift and there
is no point in waisting one's time installing and configuring NTP.

These un-synced machines are the larger part of machines out there.

They are "isolated" machines.

YET these machines may USE timestamps a LOT.

Including doing "insane" things like:

1. making a CORBA call that retrieves a timestamp in 32-bit.
2. making a SOAP call that uploads that timestamp in Ascii.

The conversion function MUST be consistent regardless of
whether the sys admin has forgotton to install NTP or
other leap-second-table-o-sync-o-software.

You cannot change POSIX from being -

A. a system that NEVER required the leap second table to be
downloaded automatically on 100's of millions of machines,

into one which:

B. now DOES require the leap second table to be downloaded
automatically on 100's of millions of machines, or else your
time conversion functions on all those machines will give
WRONG results.

-paul