On 2015-03-08 05:00 PM, Warner Losh wrote:
On Mar 8, 2015, at 10:24 AM, Brooks Harris <bro...@edlmax.com> wrote:
I think the only way the industry can eventually converge on reliable "civil time" representation
is to refine the underlying time mechanisms in POSIX in some manner that allows a migration to a more
comprehensive UTC implementation. I think if a new new POSIX time specification were to take shape it would
add an option to the the conversation at ITU-R - instead of simply "to kill Leap Seconds or not"
they'd also have "a viable migration path to comprehensive UTC timekeeping implementation" to
consider.
I think you vastly underestimate the amount of effort this change would entail.
Hi Warner,
Its a world-wide societal problem. I think it would take at least a
decade for a good specification to find its way into all main stream
systems and applications, and that's after it was agreed upon. I'd say
it can't be done, which I don't think that is underestimating it. But
never say never.
We'd trade
an off by 1 second every other year problem for an off by 36s problem in many
different
places.
Even if we left time_t as a legacy thing,
I think you have to to provide reverse compatibility for an extended
transition period. I mean an *extended* transition period, like, until
all vestiges of the legacy implementations go offline. Even then, there
will be boat loads of archival timestamps created by the legacy systems.
and defined a whole new set of interfaces that
did things pedantically correct,
I think that's what's required.
that's a lot of effort to code and adopt.
Oh, lots and lots and lots :-)
And there's almost
no economic incentive to drive the change.
I'm not sure that's true. The "dangers" of incorrect time stamps are
recognised in many quarters, not the least in financial and high-speed
trading. The possible catastrophic system failures are exactly where the
whole "kill Leap Seconds" thing comes from, where Google and others have
spent boat loads of energy and money to paper over the risk with Google
Smear. It seems to me like any and all players in the computer industry
would significantly benefit from real precision timekeeping. All
industries, jurisdictions, and citizens everywhere would benefit too.
Tons of money was spent on the "Y2K" problem. The "Leap Second" problem
is more subtle, but similar, and ubiquitous. There is not yet any
panicked demand for it, but there is building recognition of the issue.
The resources might still be mustered, especially behind a coherent and
credible design and plan.
You'd need to change the "It's just a second, who cares?" attitude before
meaningful
progress can be made in getting leap seconds right.
There are an awful lot of people that recognize the problem and do care.
Will it coalesce into a movement? Only if a good plan emerges from
somewhere. Right now everybody just runs for cover...
-Brooks
Warner
_______________________________________________
LEAPSECS mailing list
LEAPSECS@leapsecond.com
https://pairlist6.pair.net/mailman/listinfo/leapsecs
_______________________________________________
LEAPSECS mailing list
LEAPSECS@leapsecond.com
https://pairlist6.pair.net/mailman/listinfo/leapsecs
