[LEAPSECS] Bloomberg announced its smear

Steve Summit scs+ls at eskimo.com
Tue Sep 27 12:30:37 EDT 2016


Bjørn Hansen wrote:
> On Sep 27, 2016, at 23:44, Stephen Colebourne <scolebourne at joda.org> wrote:
>> But, since they are NOT going to be eliminated now, isn't it time this
>> list focussed its energies on practical solutions such as smearing?
>
> Smearing is difficult (impossible?) in any application where you get
> time signals from multiple sources or are expected to be in sync with
> multiple external only-coordinated-through-UTC systems.

It depends on where you do the smearing.  In a scheme I'm working
on, the kernel clock keeps true (non-smeared, non-Posix) UTC time
internally, but deals in smeared time on legacy Posix interfaces.
But there are also ways of getting and setting true-UTC time, for
applications that care, and this would naturally include those
(like NTP) that are responsible for time synchronization with
other systems.


More information about the LEAPSECS mailing list