I was watching the event on an Apple iPad with an app that was
displaying the ntp time together with the difference from Apple time. At
the leap second, the ntp app showed 23:59:60 and when I noticed (seconds
later, as it coincided with the glasses chinking and the fireworks of
New Year), Apple time was one second out. Apple time remained 1 second
out at least until I went to bed c 0100.
Somewhere around 00:15, I used the app to sync against my GPS-derived
Raspberry Pi ntp (actually NTPsec) and that also showed Apple time to be
1 second out.
When I next looked around 09:00, the Apple iPad time had corrected
itself, no difference from ntp.
I've not looked further, not worked out how to examine iPad logs but my
theory post-celebrating last night was that the iPad was not using ntp
"properly" but perhaps just corrects the time occasionally perhaps
hourly. Obviously that is useful for saving battery oomph and
expensive GSM bandwidth, so understandable.
All good clean fun.
Garry
Fellow time-nuts,
Several friends observed that their Apple devices, laptop and phones,
did not handle the leap second timely. One recorded the UTC+1h time with
the sequence
00:59:57
00:59:58
00:59:59
01:00:00
01:00:01
01:00:01 <- leap second inserted
01:00:02
01:00:03
The other dug up that Cocoa claims that using NTP prohibits it from
implementing it correctly, which doesn't match my laptops observation
being fed only from NTP.
Anyone else saw something similar?
The hurdles of precision time in a POSIX-damaged world...
Cheers,
Magnus
I was watching the event on an Apple iPad with an app that was
displaying the ntp time together with the difference from Apple time. At
the leap second, the ntp app showed 23:59:60 and when I noticed (seconds
later, as it coincided with the glasses chinking and the fireworks of
New Year), Apple time was one second out. Apple time remained 1 second
out at least until I went to bed c 0100.
Somewhere around 00:15, I used the app to sync against my GPS-derived
Raspberry Pi ntp (actually NTPsec) and that also showed Apple time to be
1 second out.
When I next looked around 09:00, the Apple iPad time had corrected
itself, no difference from ntp.
I've not looked further, not worked out how to examine iPad logs but my
theory post-celebrating last night was that the iPad was not using ntp
"properly" but perhaps just corrects the time occasionally perhaps
hourly. Obviously that is useful for saving battery oomph and
expensive GSM bandwidth, so understandable.
All good clean fun.
Garry
> Fellow time-nuts,
>
> Several friends observed that their Apple devices, laptop and phones,
> did not handle the leap second timely. One recorded the UTC+1h time with
> the sequence
>
> 00:59:57
> 00:59:58
> 00:59:59
> 01:00:00
> 01:00:01
> 01:00:01 <- leap second inserted
> 01:00:02
> 01:00:03
>
> The other dug up that Cocoa claims that using NTP prohibits it from
> implementing it correctly, which doesn't match my laptops observation
> being fed only from NTP.
>
> Anyone else saw something similar?
>
> The hurdles of precision time in a POSIX-damaged world...
>
> Cheers,
> Magnus
