--enable-leap-smear

Richard Laager rlaager at wiktel.com
Fri Mar 30 23:53:52 UTC 2018


On 03/20/2018 02:43 PM, Richard Laager via devel wrote:
>> 2) Is it noon to noon UTC?
> I'm still concerned about this one. In the same function, we have this:
> 
> leap_smear.interval = leap_smear_intv;
> leap_smear.intv_end = lsdata.ttime;
> leap_smear.intv_start = leap_smear.intv_end - leap_smear.interval;
> 
> Is this smearing over the interval (e.g. 24 hours) _before_ the leap
> second, as opposed to _centered on_ the leap second (which is what
> Google does)?

I confirmed that it is _before_ the leap second, rather than _centered
on_ the leap second:
https://gitlab.com/NTPsec/ntpsec/issues/475

I also found that the Python ntpq does not output leapsmearinterval or
leapsmearoffset as documented. Either the script should be updated or
the documentation should:
https://gitlab.com/NTPsec/ntpsec/issues/476

Also, NTPsec fails to accept leapsmearinterval option, for lack of
braces around a value sanity check. I fixed that an a bunch of
mixed/inconsistent whitespace here:
https://gitlab.com/NTPsec/ntpsec/merge_requests/671

This concludes my leap smear testing. Unless/until this matches the
Google proposed standard smear, I'm very hesitant to enable it in the
Debian package.

-- 
Richard


More information about the devel mailing list