Re: [LEAPSECS] UTC is doomed

From: Ed Davies <ls_at_EDAVIES.NILDRAM.CO.UK>
Date: Thu, 24 Apr 2003 13:38:49 +0100

In reply to Steve Allen I wrote:
> As I understand your argument it is that because leap seconds will
> eventually become unworkable any unrelated reasons given for getting
> rid of them in the short term are somehow weakened or invalidated.
> Can this really be what you mean?

Steve Allen replied:
> No. I mean that there is no impending operational flaw in the
> definition of UTC which we have used since 1972. Any change in UTC
> must be motivated by other, broadly, justifiably, urgent reasons.
> ...

Indeed, but, as far as I know, nobody is suggesting that there is
an impending operational flaw in the definition of UTC.

I think that the calls for change to UTC are motivated by the
"justifiable and urgent" reason that computer and similar systems
have had a dreadfully bad history of dealing with discontinuities
in timescales and that, as more systems are based on to the second
or tighter levels of synchronization, the "surprise" arrival of
leap seconds is going to cause trouble. Or, to put it another way,
dealing with leap seconds properly is a nuisance to implement and,
worse, difficult to test.

The question is, does the "cost" of implementing timescales properly
in all the computer based systems affected outweigh the "costs" of
having UTC drift away from the various flavours of UT?

I think not - that is, I think that if leap seconds are kept they
can be dealt with in computer systems perfectly adequately with
only a bit of education followed by a bit more careful design and
coding.

However, I do think that now that there is discussion of getting
rid of leap seconds we should come to a conclusion, one way or the
other, pretty quickly so that people don't think "oh well, I'll
not bother with leap seconds in this system because they'll
probably go away soon, anyway".

In this sense I do disagree with Steve: I think there really is a
rush to come to a conclusion, even though, on balance, I'd rather
that conclusion was to continue leap seconds. In other words,
I hope we rush to stay still :-)

Ed Davies.
Received on Thu Apr 24 2003 - 07:54:15 PDT